TokenScale checks each provider source every night and records what the check could reproduce. A successful observation is verified; otherwise the last known value is carried forward and marked. This page audits every missed or partial check. Provider pages remain authoritative for current published prices; TokenScale preserves the independent dated record of what it observed.
A fully verified night means a human-readable price was read from the provider's own pricing page that night, for every tracked tier. A carried-forward night means the check did not complete and the previous price was held. A partly verified night means some tiers were read and the rest were held: it is counted as a miss for the tiers it missed, and the count is stated below. Those days are drawn in grey on every chart, and they stay grey: a later check confirms only its own date, it cannot reach back and confirm one it missed. The gap is part of the record.
Disclosure is the commitment here, not perfection. The verification rate will never be 100%, because the checks run on real machines in the real world. The commitment is that every night the record missed appears on this page, with the reason.
The site launched on 18 May 2026; the first automated nightly point landed on 19 May. Coverage grew with the site: 16 providers from 19 May, 21 from 2 June, and all 22 from 26 June, when Sakana AI joined the board.
| Nights | Run | What happened | Disclosed |
|---|---|---|---|
| 18 Nov | opening desk | The 8 PM opening desk freshly reopened all 22 official routes and accounted for all 66 Standard tiers at 56 verified plus ten visibly carried: Mistral Mid/Pro, Groq Lite/Mid/Pro, Perplexity Lite/Mid/Pro, MiniMax Lite and OpenRouter Pro. Batch remains 21 supported, three unsupported and 42 unestablished, with three blocked Groq exact checks. Cache activation remains 5/3/56/2 and observed evidence 40/3/18/5 across all 22 provider routes and all 66 exact records; all five supported routes reproduced and the complete field diff is empty. All 27 continuing flags were revisited. Google launched generally available Gemini 4 Neon at $3/$15 with $0.15 cached input, but its first-party pages do not uniquely map it to a tracked slot, so no lineup move is inferred. Will owns the Neon decision, Groq access and the other carried lineup decisions; the independent 10 PM close remains mandatory. | This page, 18 Nov 2026 |
| 17 Nov | 2 desks, closed | The 8 PM opening desk and independent 10 PM close each freshly reopened all 22 official routes and accounted for all 66 Standard tiers at 56 verified plus ten visibly carried: Mistral Mid/Pro, Groq Lite/Mid/Pro, Perplexity Lite/Mid/Pro, MiniMax Lite and OpenRouter Pro. Batch closes at 21 supported, three unsupported and 42 unestablished, with three blocked Groq exact checks. Cache activation closes at 5/3/56/2 and observed evidence at 40/3/18/5 across all 22 provider routes and all 66 exact records; all five supported routes reproduced and the complete field diff is empty. The close independently reconfirmed OpenAI's GPT-6.2 Astra and GPT-6.2 Sol launch plus the four-item source-watch flush without a tracked amount, model, Batch rule, Cache field or Journal-premise change. All 27 unresolved flags were revisited. Will owns Groq access and every ambiguous lineup decision. | This page, 17 Nov 2026 |
| 16 Nov | 1 night, 2 desks, closed | The 8 PM opening desk and independent 10 PM close each freshly reopened all 22 official routes. The close accounts for all 66 Standard tiers at 56 verified plus ten visibly carried: Groq Lite/Mid/Pro, Perplexity Lite/Mid/Pro, Mistral Medium 3.5 and Large 3, MiniMax M2 and OpenRouter DeepSeek R1. Batch closes at 21 supported, three unsupported and 42 unestablished with three blocked Groq checks. Cache activation closes at five supported, three explicitly unavailable, 56 unestablished and two blocked; observed evidence closes at 40 priced, three unavailable, 18 unestablished and five blocked. All five supported Cache routes reproduced and the complete 66-record field diff is empty. Alibaba launched generally available Qwen4-Max, but published no unique map to tracked Qwen3.7-Max. Mistral Large 4 is now generally available at $0.70 input, $0.07 cached input and $2.10 output, but one successor still maps to two tracked roles. No tracked amount, model, Batch rule or Cache field changed; Will owns Groq access and the held lineup decisions. | This page, 16 Nov 2026 |
| 15 Nov | 1 night, 2 desks, closed | The 8 PM opening desk and independent 10 PM close each freshly reopened all 22 official routes. The close accounts for all 66 Standard tiers at 56 verified plus ten visibly carried: Groq Lite/Mid/Pro, Perplexity Lite/Mid/Pro, Mistral Medium 3.5 and Large 3, MiniMax M2 and OpenRouter DeepSeek R1. Both focused rereads reproduced every carry reason; all six Groq Standard, Batch, Cache and lifecycle routes remained blocked by identical HTTP 403 responses. Batch closes at 21 supported, three unsupported and 42 unestablished with three blocked exact Groq checks. Cache activation closes at five supported, three explicitly unavailable, 56 unestablished and two blocked; observed evidence closes at 40 priced, three unavailable, 18 unestablished and five blocked. All five supported Cache routes reproduced and the complete 66-record field diff is empty. All 26 continuing flags and the opening visitor-facing claims were rechecked. GPT-5.6 Sol's promotion through November 30 and Gemini 4 Hydrogen's factual launch remain supported. No tracked price, model, Batch rule, Cache field, qualified successor, large-change classification or Journal premise changed. Will owns Groq access and the held Perplexity, Mistral, MiniMax and OpenRouter decisions. | This page, 15 Nov 2026 |
| 14 Nov | 1 night, recovered closing evidence | The same-day sealed independent close at 10 PM was recovered on 15 November after the intro-film build blocker was removed. Its complete receipt covers all 22 official routes and all 66 Standard tiers at 56 verified plus ten visibly carried: Groq Lite/Mid/Pro, Perplexity Lite/Mid/Pro, Mistral Medium 3.5 and Large 3, MiniMax M2 and OpenRouter DeepSeek R1. Batch is 21 supported, three unsupported and 42 unestablished with three blocked exact Groq checks. Cache activation is five supported, three explicitly unavailable, 56 unestablished and two blocked; observed evidence is 40 priced, three unavailable, 18 unestablished and five blocked. All five supported Cache routes reproduced and the complete field diff is empty. The 66 dated history points, same-night ledger rows and closing receipt were restored without replacing the later intro-film retirement. No tracked price, model, Batch rule or Cache field changed. November 11 to 13 are preserved separately as fully carried grey continuity nights; none is claimed as verified. | This page, 15 Nov 2026 |
| 11–13 Nov | 3 nights, 0 verified / 66 carried each | The official-source desk did not record a complete same-day pass or canonical closing receipt on 11, 12 or 13 November. All 66 tiers were explicitly carried on each date as visibly grey observations copied from the last recorded values, so the timeline does not skip the missed nights. None of those 198 observations is claimed as verified, and no Batch or Cache check is backdated. On 15 November the release blockage was traced to intro-film exports rewriting public assets after packaging; the film feature and renderer were retired, and browser-control failures were given an explicit non-terminal fallback so the guarded nightly path can resume. | This page, 15 Nov 2026 |
| 10 Nov | 1 night, 2 desks, closed | The 8 PM opening desk and independent 10 PM close each freshly reopened all 22 official routes. The close accounts for all 66 Standard tiers at 56 verified plus ten visibly carried: Groq Lite/Mid/Pro, Perplexity Lite/Mid/Pro, Mistral Medium 3.5 and Large 3, MiniMax M2 and OpenRouter DeepSeek R1. The complete pass and two focused rereads reproduced every carry reason; all six Groq Standard, Batch, Cache and lifecycle routes remained blocked by identical HTTP 403 responses. Batch closes at 21 supported, three unsupported and 42 unestablished with three blocked exact Groq checks. Cache activation closes at five supported, three explicitly unavailable, 56 unestablished and two blocked; observed evidence closes at 40 priced, three unavailable, 18 unestablished and five blocked. All five supported Cache routes reproduced and the complete 66-record field diff is empty. All 22 continuing flags and the opening visitor-facing claims were rechecked. GPT Image 2 lifecycle and Qwen4-Max-Preview remain two untracked Class B queue items below threshold. No tracked price, model, Batch rule, Cache field, qualified successor, large-change classification or Journal premise changed. Will owns Groq access and the held Perplexity, Mistral, MiniMax and OpenRouter decisions. | This page, 10 Nov 2026 |
| 9 Nov | missed night | No same-day official-source pass or canonical receipt exists for November 9. All 66 Standard values are preserved as visibly carried, with no read backdated from November 10. Batch and Cache evidence remains dated to the last actual read. This is a continuity incident, not a verified night. | This page, 10 Nov 2026 |
| 8 Nov | 1 night, opening watchdog recovery | The missing 8 PM branch was recovered before the independent close from fresh reads of all 22 official routes. Standard accounts for all 66 tiers at 56 verified plus ten visibly carried: Groq Lite/Mid/Pro, Perplexity Lite/Mid/Pro, Mistral Medium 3.5 and Large 3, MiniMax M2 and OpenRouter DeepSeek R1. Two focused rereads reproduced every hold. Batch is 21 supported, three unsupported and 42 unestablished with three blocked Groq exact checks. Cache activation is five supported, three explicitly unavailable, 56 unestablished and two blocked; observed evidence is 40 priced, three unavailable, 18 unestablished and five blocked. All five supported Cache routes reproduced and the complete field diff is empty. The dated record starts Sakana Namazu and Fugu Max in Will-authorized slots and Meta Muse Spark 1.3 at its unchanged rate, preserving all predecessor history. The independent 10 PM close remains mandatory. | This page, 8 Nov 2026 |
| 7 Nov | 1 night, 2 desks, closed | The 8 PM opening desk and independent close each freshly reopened all 22 official routes. The independent close accounted for all 66 Standard tiers at 56 verified plus ten visibly carried: Groq Lite, Mid and Pro, Perplexity Lite, Mid and Pro, Mistral Medium 3.5 and Large 3, MiniMax M2 and OpenRouter DeepSeek R1. The complete pass and three focused rereads left every Groq Standard, Batch, Cache and lifecycle route blocked by identical HTTP 403 responses and reproduced every other carry reason. Batch closes at 21 supported, three unsupported and 42 unestablished with three blocked exact Groq checks. Cache activation closes at five supported, three explicitly unavailable, 56 unestablished and two blocked; observed evidence closes at 40 priced, three unavailable, 18 unestablished and five blocked. All five supported Cache routes reproduced and the complete 66-record field diff is empty. All 23 continuing flags and the opening visitor-facing claims were rechecked. GPT-6-Cyber remains one untracked Class B queue item below threshold. No tracked price, model, Batch rule, Cache field, qualified successor, large-change classification or Journal premise changed. Will owns Groq access and the held Perplexity, Mistral, MiniMax and OpenRouter decisions. | This page, 7 Nov 2026 |
| 6 Nov | 1 night, 2 desks, closed | The 8 PM opening desk and independent close each freshly reopened all 22 official routes. The independent close accounted for all 66 Standard tiers at 59 verified plus seven visibly carried: Perplexity Lite, Mid and Pro, Mistral Medium 3.5 and Large 3, MiniMax M2 and OpenRouter DeepSeek R1. Two focused rereads reproduced every carry reason. Batch closes at 21 supported, three unsupported and 42 unestablished with zero blocked exact checks. Cache activation closes at five supported, three explicitly unavailable, 56 unestablished and two blocked; observed evidence closes at 43 priced, three unavailable, 18 unestablished and two blocked. All five supported Cache routes reproduced and the complete 66-record field diff is empty. All 23 due flags and the opening visitor-facing claims were rechecked. GPT-6-Cyber remains one untracked Class B queue item below threshold. No tracked price, model, Batch rule, Cache field, qualified successor, large-change classification or Journal premise changed. Will owns the Perplexity, Mistral, MiniMax and OpenRouter decisions. | This page, 6 Nov 2026 |
| 5 Nov | 1 night, scheduled desks missed; morning recovery | The November 5 8 PM and 10 PM official-source desks did not leave branch evidence, and the same-night watchdog did not run. There was no independent close before the Pacific date ended, so all 66 Standard tiers are permanently carried forward from November 4 and shown grey. The delayed watchdog reopened all 22 provider routes on November 6, accounted for all 66 current Batch records and all 22 Cache routes plus 66 exact Cache records, and ran two focused rereads of the 12 current blocker routes. Those late reads are current follow-up evidence only and do not backdate November 5 verification. OpenAI GPT-6-Cyber entered the untracked Class B source-watch queue. No tracked amount, model, Batch rule, Cache field or lineup changed. | This page, 6 Nov 2026 |
| 4 Nov | 1 night, 2 desks, closed | The 8 PM opening desk and independent close each freshly reopened all 22 official routes. The independent close accounted for all 66 Standard tiers at 59 verified plus seven visibly carried: Perplexity Lite, Mid and Pro, Mistral Medium 3.5 and Large 3, MiniMax M2 and OpenRouter DeepSeek R1. Two focused rereads reproduced every carry reason. Batch closes at 21 supported, three unsupported and 42 unestablished with zero blocked exact checks. Cache activation closes at five supported, three explicitly unavailable, 56 unestablished and two blocked; observed evidence closes at 43 priced, three unavailable, 18 unestablished and two blocked. All five supported Cache routes reproduced and the complete 66-record field diff is empty. All 23 due flags and the opening visitor-facing claims were rechecked. The opening five-item Class B source-watch cluster remains supported. No tracked price, model, Batch rule, Cache field, qualified successor, large-change classification or Journal premise changed. Will owns the Perplexity, Mistral, MiniMax and OpenRouter decisions. | This page, 4 Nov 2026 |
| 3 Nov | 1 night, 2 desks, closed | The 8 PM opening desk and independent close each freshly reopened all 22 official routes. The independent close accounted for all 66 Standard tiers at 59 verified plus seven visibly carried: Perplexity Lite, Mid and Pro, Mistral Medium 3.5 and Large 3, MiniMax M2 and OpenRouter DeepSeek R1. Two focused rereads reproduced every carry reason. Batch closes at 21 supported, three unsupported and 42 unestablished with zero blocked exact checks. Cache activation closes at five supported, three explicitly unavailable, 56 unestablished and two blocked; observed evidence closes at 43 priced, three unavailable, 18 unestablished and two blocked. All five supported Cache routes reproduced and the complete 66-record field diff is empty. All 23 due flags and the opening visitor-facing claims were rechecked. The opening three-item Class B source-watch cluster remains supported. No tracked price, model, Batch rule, Cache field, qualified successor, large-change classification or Journal premise changed. Will owns the Perplexity, Mistral, MiniMax and OpenRouter decisions. | This page, 3 Nov 2026 |
| 2 Nov | 1 night, 2 desks, closed | The 8 PM opening desk and independent close each freshly reopened all 22 official routes. The independent close accounted for all 66 Standard tiers at 59 verified plus seven visibly carried: Perplexity Lite, Mid and Pro, Mistral Medium 3.5 and Large 3, MiniMax M2 and OpenRouter DeepSeek R1. Two focused rereads reproduced every carry reason. Batch closes at 21 supported, three unsupported and 42 unestablished with zero blocked exact checks. Cache activation closes at five supported, three explicitly unavailable, 56 unestablished and two blocked; observed evidence closes at 43 priced, three unavailable, 18 unestablished and two blocked. All five supported Cache routes reproduced and the complete 66-record field diff is empty. All 23 due flags and the opening visitor-facing claims were rechecked. The two-item Class B queue remains below threshold. No tracked price, model, Batch rule, Cache field, qualified successor, large-change classification or Journal premise changed. Will owns the Perplexity, Mistral, MiniMax and OpenRouter decisions. | This page, 2 Nov 2026 |
| 1 Nov | 1 night, 2 desks, closed | The 8 PM opening desk and independent close each freshly reopened all 22 official routes. The closing pass accounted for all 66 Standard tiers at 59 verified plus seven visibly carried: Perplexity Lite, Mid and Pro, Mistral Medium 3.5 and Large 3, MiniMax M2 and OpenRouter DeepSeek R1. Two focused closing rereads reproduced every carry reason. Batch closes at 21 supported, three unsupported and 42 unestablished with zero blocked exact checks. Cache activation closes at five supported, three explicitly unavailable, 56 unestablished and two blocked; observed evidence is 43 priced, three unavailable, 18 unestablished and two blocked. All five supported Cache routes reproduced and the complete 66-record field diff is empty. One untracked xAI image-model retirement entered the Class B source-watch queue. No tracked price, model, Batch rule, Cache field, qualified successor, large-change classification or Journal premise changed. Will owns the Perplexity, Mistral, MiniMax and OpenRouter decisions. | This page, 1 Nov 2026 |
| 31 Oct | 1 night, 2 desks, closed | The 8 PM opening desk and independent close each freshly reopened all 22 official routes. The closing pass accounted for all 66 Standard tiers at 59 verified plus seven visibly carried: Mistral Medium 3.5 and Large 3, all three Perplexity tiers, MiniMax M2 and OpenRouter DeepSeek R1. Two focused closing rereads reproduced every carry reason. Batch closes at 21 supported, three unsupported and 42 unestablished with zero blocked exact checks. Cache activation closes at five supported, three explicitly unavailable, 56 unestablished and two blocked; observed evidence is 43 priced, three unavailable, 18 unestablished and two blocked. All five activated routes reproduced and the complete 66-record field diff is empty. Groq remains recovered at unchanged Standard and Batch rates; its observed Cache remains noncalculable because the exact per-model minimum is unpublished. No closing Standard amount, model, Batch rule, Cache field, qualified successor, large-change classification, launch or Journal premise changed. Will owns the Perplexity, Mistral, MiniMax and OpenRouter decisions. | This page, 31 Oct 2026 |
| 30 Oct | 1 night, opening partial + watchdog recovery | The 8 PM desk freshly reopened all 22 official routes; the 9:10 watchdog retried only the six recorded blockers. Two fresh canonical Kimi pricing reads plus the distinct current model catalog restored K2.6 Lite and Mid to verified at unchanged $0.95/$4, leaving all 66 Standard tiers at 56 verified plus ten visibly carried. Exact Kimi Batch rows restore the registry to 21 supported, three unsupported and 42 unestablished, with the three Groq exact checks blocked. Cache activation remains five supported, three explicitly unavailable, 53 unestablished and five blocked; observed evidence is 40 priced, three unavailable, 18 unestablished and five blocked. Mistral Medium 3.5 and Large 3 remain held because named successor Large 4 is Public Preview and maps to two tracked roles. Three first-party facts crossed the Class B source-watch threshold. All 24 due flags were rechecked. Will owns the ambiguous lineup decisions, and the independent 10 PM close is still required. | This page, 30 Oct 2026 |
| 29 Oct | 1 night, 2 desks, closed | The 8 PM opening and independent 10 PM closing official-source desks each freshly reopened all 22 routes. The independent close accounted for all 66 Standard tiers at 56 verified plus ten visibly carried: Groq Lite/Mid/Pro, Perplexity Lite/Mid/Pro, Kimi K2.6 Lite/Mid, MiniMax M2 and OpenRouter DeepSeek R1. Two focused closing rereads reproduced every carry reason. Batch closes at 20 supported, four unsupported and 42 unestablished, with the three Groq exact checks blocked. Cache activation closes at five supported, three explicitly unavailable, 53 unestablished and five blocked; observed evidence is 39 priced, three explicitly unavailable, 19 unestablished and five blocked. All five activated routes reproduced and the complete 66-record field diff is empty. The opening four-item Class B queue was recorded and cleared; no new closing item changed a tracked price, model, Batch rule, Cache field or Journal premise. All 24 due flags were rechecked. Will owns Groq access and every ambiguous lineup or service-basis decision. | This page, 29 Oct 2026 |
| 28 Oct | 1 night, 2 desks, closed | The 8 PM opening and independent 10 PM closing official-source desks each freshly reopened all 22 routes. The independent close accounted for all 66 Standard tiers at 56 verified plus ten visibly carried: Groq Lite/Mid/Pro, Perplexity Lite/Mid/Pro, Kimi K2.6 Lite/Mid, MiniMax M2 and OpenRouter DeepSeek R1. Two focused closing rereads reproduced every carry reason. Batch closes at 20 supported, four unsupported and 42 unestablished, with the three Groq exact checks blocked. Cache activation closes at five supported, three explicitly unavailable, 53 unestablished and five blocked; observed evidence is 39 priced, three explicitly unavailable, 19 unestablished and five blocked. All five activated routes reproduced and the complete 66-record field diff is empty. OpenAI's subscriber API-credit benefit and xAI custom moderation controls remain a two-item Class B queue without changing a tracked price, model, Batch rule, Cache field or Journal premise. All 24 due flags were rechecked. Will owns Groq access and every ambiguous lineup or service-basis decision. | This page, 28 Oct 2026 |
| 27 Oct | 1 night, 2 desks, closed | The 8 PM opening and independent 10 PM closing official-source desks each freshly reopened all 22 routes. The independent close accounted for all 66 Standard tiers at 56 verified plus ten visibly carried: Groq Lite/Mid/Pro, Perplexity Lite/Mid/Pro, Kimi K2.6 Lite/Mid, MiniMax M2 and OpenRouter DeepSeek R1. Two focused closing rereads reproduced every carry reason. Batch closes at 20 supported, four unsupported and 42 unestablished, with the three Groq exact checks blocked. Cache activation closes at five supported, three explicitly unavailable, 53 unestablished and five blocked; observed evidence is 39 priced, three explicitly unavailable, 19 unestablished and five blocked. All five activated routes reproduced and the complete 66-record field diff is empty. xAI custom moderation controls entered the one-item Class B queue without changing tracked data. No tracked price, model, Batch rule, Cache field, qualified successor, large change, Class A launch or Journal premise changed. Will owns Groq access and every ambiguous lineup decision. | This page, 27 Oct 2026 |
| 26 Oct | 1 night, 2 desks, closed | The 9:10 watchdog recovered the missing 8 PM opening branch; the independent close at 10 PM then repeated fresh cache-busted reads across all 22 official routes. It accounted for all 66 Standard tiers at 56 verified plus ten visibly carried: Groq Lite/Mid/Pro, Perplexity Lite/Mid/Pro, Kimi K2.6 Lite/Mid, MiniMax M2 and OpenRouter DeepSeek R1. The carry reasons reproduced in two focused rereads. Batch is 20 supported, four unsupported and 42 unestablished, with the three Groq exact checks blocked. Cache activation is five supported, three explicitly unavailable, 53 unestablished and five blocked; observed evidence is 39 priced, three explicitly unavailable, 19 unestablished and five blocked. All five activated routes reproduced and the complete 66-record field diff is empty. OpenAI launched GPT-6.1 Terra, but published no unique successor mapping from tracked GPT-5.6 Terra, so no lineup change was inferred. Will owns Groq access and every ambiguous lineup or service-basis decision. | This page, 26 Oct 2026 |
| 25 Oct | 2 desks, closed partial | Both official-source desks freshly attempted all 22 routes; the independent close verified 56 of 66 Standard tiers and left ten visibly carried. Groq Lite, Mid and Pro remain carried because the current public models and Batch documentation returned byte-identical HTTP 403 responses on the closing read and two focused rereads. Perplexity Lite, Mid and Pro, Kimi K2.6 Lite and Mid, MiniMax M2 and OpenRouter DeepSeek R1 remain carried because their current exact official surfaces do not reproduce the complete tracked rows. Kimi K3 remains freshly verified at $3 input, $15 output and $0.30 cached input. This is a source-access and exact-row verification event, not a price move, withdrawal or retirement claim, and no successor is inferred. Batch closes at 20 supported, four unsupported and 42 unestablished, with the three Groq exact checks blocked. Cache activation closes at five supported, three explicitly unavailable, 53 unestablished and five blocked, with observed evidence 39 priced, three explicitly unavailable, 19 unestablished and five blocked. All five activated Cache routes reproduced and the complete 66-route field diff is empty. Will owns the Groq access escalation and every ambiguous lineup or service-basis decision. | This page, 25 Oct 2026 |
| 24 Oct | 2 desks, closed partial | Both official-source desks freshly attempted all 22 routes; the independent close verified 56 of 66 Standard tiers and left ten visibly carried. Groq Lite, Mid and Pro remain carried because the current public models and Batch documentation returned HTTP 403 on the closing read and two independent focused rereads. Perplexity Lite, Mid and Pro, Kimi K2.6 Lite and Mid, MiniMax M2 and OpenRouter DeepSeek R1 remain carried because their current exact official surfaces do not reproduce the complete tracked rows. Kimi K3 remains freshly verified at $3 input, $15 output and $0.30 cached input. This is a source-access and exact-row verification event, not a price move, withdrawal or retirement claim, and no successor is inferred. Batch closes at 20 supported, four unsupported and 42 unestablished, with the three Groq checks blocked. Cache activation closes at five supported, three explicitly unavailable, 53 unestablished and five blocked, with observed evidence 39 priced, three explicitly unavailable, 19 unestablished and five blocked. All five activated Cache routes reproduced and the complete 66-route field diff is empty. Will owns the Groq access escalation and every ambiguous lineup or service-basis decision. | This page, 24 Oct 2026 |
| 23 Oct | 2 desks, closed partial | Both official-source desks freshly attempted all 22 routes; the independent close verified 56 of 66 Standard tiers and left ten visibly carried. Perplexity Lite, Mid and Pro, Kimi K2.6 Lite and Mid, MiniMax M2 and OpenRouter DeepSeek R1 remain carried because their current exact official surfaces do not reproduce the complete tracked rows. Kimi's public platform surface again reproduced unchanged tracked K3 at $3 input, $15 output and $0.30 cached input. Groq's current models and Batch documentation both returned HTTP 403 on repeated cache-busted closing reads after the opening desk had reproduced the tracked rows, so GPT-OSS 20B and both GPT-OSS 120B tier observations are now grey at their unchanged held values. This is a source-access event, not a price move, withdrawal or retirement claim. Batch registry status remains 20 supported, four unsupported and 42 unestablished, but the three Groq rules are closing blockers and no green close is claimed. Cache activation closes at five supported, three explicitly unavailable, 53 unestablished and five blocked, with observed evidence 39 priced, three explicitly unavailable, 19 unestablished and five blocked. Kimi K3.1 and the K2 lifecycle notice remain a held lineup decision; no successor is inferred and Will owns all editorial choices. | This page, 23 Oct 2026 |
| 22 Oct | 2 desks, closed | The 8 PM opening and independent 10 PM closing official-source desks each freshly covered all 22 routes. The independent close verified 58 of 66 Standard tiers. Perplexity Lite, Mid and Pro, Kimi Lite, Mid and Pro, MiniMax M2 and OpenRouter DeepSeek R1 remain visibly carried because their current exact official surfaces do not reproduce the complete tracked rows; two byte-identical focused closing reads reproduced every hold, and no retirement or successor is inferred. Batch closes at 20 supported, four unsupported and 42 unestablished. Cache closes at five supported, three explicitly unavailable, 56 unestablished and two blocked across all 66 exact records and 22 provider routes, with observed evidence 41 priced, three explicitly unavailable, 20 unestablished and two blocked. Nine same-date Class B platform and catalog facts were recorded as one source-watch cluster without changing any tracked amount, model, Batch rule or Cache field. | This page, 22 Oct 2026 |
| 21 Oct | 2 desks, closed | The 8 PM opening and independent 10 PM closing official-source desks each freshly covered all 22 routes. The independent close verified 58 of 66 Standard tiers. Perplexity Lite, Mid and Pro remain visibly carried because current Agent API routes still publish one Sonar basis without the exact context and distinct availability needed to preserve three tracked slots. Kimi Lite, Mid and Pro remain carried because the current Standard and Batch price tables are still empty while K2.6 and K3 remain listed. MiniMax M2 and OpenRouter DeepSeek R1 also remain carried because their exact current source surfaces omit those tracked models; their Cache routes remain blocked. No tracked Standard amount, model label, exact Batch rule or Cache field changed at the close. Batch closes at 20 supported, four unsupported and 42 unestablished. Cache closes at five supported, three explicitly unavailable, 56 unestablished and two blocked across all 66 exact records and 22 provider routes, with observed evidence 41 priced, three explicitly unavailable, 20 unestablished and two blocked. Both OpenAI Cache lifetime choices reproduced unchanged. Alibaba Model Studio's new untracked GLM-5.3-Prime host/catalog row remains one Class B queue item, and all lineup or service-basis decisions remain with Will. | This page, 21 Oct 2026 |
| 20 Oct | 1 night, opening partial | The 8 PM official-source desk freshly covered all 22 routes and verified 58 of 66 Standard tiers. Perplexity Lite, Mid and Pro remain visibly carried because current Agent API routes publish one Sonar basis without the exact context and distinct availability needed to preserve three tracked slots. MiniMax M2 and OpenRouter DeepSeek R1 also remain carried because their exact current source surfaces omit those tracked models. Two byte-identical Kimi reads kept K2.6 and K3 available but returned empty Standard and Batch price tables, so all three Kimi Standard rows are carried at their last verified values, two K2.6 Batch records are unestablished and all three observed Kimi Cache amounts are unestablished. Batch is 20 supported, four unsupported and 42 unestablished. Cache activation is five supported, three explicitly unavailable, 56 unestablished and two blocked, with observed evidence 41 priced, three explicitly unavailable, 20 unestablished and two blocked. Mistral's four-model Voxtral 3 speech family was recorded as a specialist Class A launch without changing the tracked text lineup. The independent 10 PM close is still required. | This page, 20 Oct 2026 |
| 19 Oct | 2 desks, closed | The 8 PM opening and independent 10 PM closing official-source desks each freshly covered all 22 routes. The independent close verified 61 of 66 Standard tiers. Perplexity Lite, Mid and Pro remain visibly carried because current Agent API routes still publish one Sonar basis without the exact context and distinct availability needed to preserve three tracked slots. MiniMax M2 and OpenRouter DeepSeek R1 also remain carried because their exact current source surfaces omit those tracked models; their Cache routes remain blocked. No tracked amount, model label, exact Batch rule or Cache field changed at the close. Batch closes at 22 supported, four unsupported and 40 unestablished. Cache closes at five supported, three explicitly unavailable, 56 unestablished and two blocked across all 66 exact records and 22 provider routes, with observed evidence 44 priced, three explicitly unavailable, 17 unestablished and two blocked. The three-item Class B source-watch cluster remains untracked, and all lineup or service-basis decisions remain with Will. | This page, 19 Oct 2026 |
| 18 Oct | 2 desks, closed | The 8 PM opening and independent 10 PM closing official-source desks each freshly covered all 22 routes. The independent close verified 62 of 66 Standard tiers. The three Perplexity Sonar tiers remain visibly carried because current Agent API routes still publish one Sonar basis without the exact context and distinct availability needed to preserve three tracked slots. MiniMax M2 also became carried after two byte-identical closing reads no longer reproduced it in the current supported-model catalog or the rewritten prompt-cache guide. No tracked amount or model label changed. Batch closes at 22 supported, four unsupported and 40 unestablished. Cache closes at five supported, three explicitly unavailable, 57 unestablished and one blocked across all 66 exact records and 22 provider routes, with observed evidence 44 priced, three explicitly unavailable, 18 unestablished and one blocked. MiniMax M2's former Cache calculation is disabled pending Will's lineup decision. Meta Muse Spark Open remains the one-item Class B queue. | This page, 18 Oct 2026 |
| 17 Oct | 2 desks, closed | The 8 PM opening and independent 10 PM closing official-source desks each freshly covered all 22 routes. The independent close verified 63 of 66 Standard tiers; the three Perplexity Sonar tiers remain visibly carried because four current Agent API Markdown routes still publish one Sonar basis without the exact context and distinct availability needed to preserve three tracked slots. No tracked Standard amount, model, exact Batch rule or Cache field changed. Batch closes at 22 supported, four unsupported and 40 unestablished. Cache closes at six supported, three explicitly unavailable, 57 unestablished and zero blocked across all 66 exact records and 22 provider routes, with observed evidence 45 priced, three explicitly unavailable, 18 unestablished and zero blocked. The close reconfirmed DeepSeek V4.1 Flash 1017 behind the stable alias with unchanged pricing; Fireworks' current table independently corroborated the same checkpoint at $0.30/$1.20. TokenScale preserved the generic tracked label and earlier history pending Will's naming preference. Nine lineup, naming or service-basis decisions remain with Will. | This page, 17 Oct 2026 |
| 16 Oct | 2 desks, closed | The 8 PM opening and independent 10 PM closing official-source desks each freshly covered all 22 routes. The independent close verified 63 of 66 Standard tiers; the three Perplexity Sonar tiers remain visibly carried because four current Agent API routes still publish one Sonar basis without the exact context and distinct availability needed to preserve three tracked slots. No tracked Standard amount, model, exact Batch rule or Cache field changed. Batch closes at 22 supported, four unsupported and 40 unestablished. Cache closes at six supported, three explicitly unavailable, 57 unestablished and zero blocked across all 66 exact records and 22 provider routes, with observed evidence 45 priced, three explicitly unavailable, 18 unestablished and zero blocked. The close reconfirmed the seven-item Class B source-watch cluster and corrected DeepSeek's free-period wording to the newly published 1 November billing start. Eight lineup or service-basis decisions remain with Will. | This page, 16 Oct 2026 |
| 15 Oct | 2 desks, closed | The 8 PM opening and independent 10 PM closing official-source desks each freshly covered all 22 routes. The independent close verified 63 of 66 Standard tiers; the three Perplexity Sonar tiers remain visibly carried because four current Agent API routes still publish one Sonar basis without the exact context and distinct availability needed to preserve three tracked slots. No tracked Standard amount, model, exact Batch rule or Cache field changed. Batch closes at 22 supported, four unsupported and 40 unestablished. Cache closes at six supported, three explicitly unavailable, 57 unestablished and zero blocked across all 66 exact records and 22 provider routes, with observed evidence 45 priced, three explicitly unavailable, 18 unestablished and zero blocked. The close reconfirmed Mistral Timestral as a specialist Class A family launch and left Anthropic Fast mode plus OpenAI GPT-Live Decisions integration in the two-item Class B queue. Eight lineup or service-basis decisions remain with Will. | This page, 15 Oct 2026 |
| 14 Oct | 1 night, partial | Both official-source desks freshly covered all 22 routes; the independent close verified 63 of 66 Standard tiers. The three Perplexity Sonar tiers remain visibly carried because four current Agent API routes still publish one Sonar basis without the exact context and distinct availability needed to preserve three tracked slots. No tracked Standard amount, model, exact Batch rule or Cache field changed. Batch closes at 22 supported, four unsupported and 40 unestablished. Cache closes at six supported, three explicitly unavailable, 57 unestablished and zero blocked across all 66 exact records and 22 provider routes, with observed evidence 45 priced, three explicitly unavailable, 18 unestablished and zero blocked. The independent close reproduced the four-item Class B source-watch cluster: OpenAI's legacy GPT-4 deprecations, xAI Voice 3 with Voice 2 retirement, Together's Kimi K3 return to $3/$15 and Together's GLM-5.3 Flash quantization change. Seven lineup choices and Perplexity's service basis remain with Will. | This page, 14 Oct 2026 |
| 13 Oct | 1 night, partial | Both official-source desks freshly covered all 22 routes; the independent close verified 63 of 66 Standard tiers. The three Perplexity Sonar tiers remain visibly carried because current Agent API pricing and migration guidance publish one Sonar basis without the exact context and distinct availability needed to preserve three tracked slots. No Standard amount, tracked model, exact Batch rule or Cache field changed. Batch closes at 22 supported, four unsupported and 40 unestablished. Cache closes at six supported, three explicitly unavailable, 57 unestablished and zero blocked across all 66 exact records and 22 provider routes, with observed evidence 45 priced, three explicitly unavailable, 18 unestablished and zero blocked. GPT-6.1 Sol, Mistral Large 4, Grok 4.7, Qwen3.8 Max, Muse Spark 1.3 and Gemini 4 Argon remain held without qualifying unique successor maps. Claude Haiku 5.5 now has explicit Haiku 4.5 migration evidence and complete Standard, Batch and Cache terms, but the guarded application was refused, so tracked Haiku 4.5 remains unchanged pending Will. | This page, 13 Oct 2026 |
| 12 Oct | 1 night, partial | Both official-source desks freshly covered all 22 routes; the independent close verified 63 of 66 Standard tiers. The three Perplexity Sonar tiers remain visibly carried because current Agent API pricing and migration guidance publish one Sonar basis without the exact context and distinct availability needed to preserve three tracked slots. No Standard amount, tracked model, exact Batch rule or Cache field changed. Batch closes at 22 supported, four unsupported and 40 unestablished. Cache closes at six supported, three explicitly unavailable, 57 unestablished and zero blocked across all 66 exact records and 22 provider routes, with observed evidence 45 priced, three explicitly unavailable, 18 unestablished and zero blocked. GPT-6.1 Sol, Claude Haiku 5.5, Mistral Large 4, Grok 4.7, Qwen3.8 Max and Muse Spark 1.3 remain held without qualifying successor maps. GPT-6.1 Sol Ultrafast, Qwen Audio 3.1 TTS Flash and Meta Muse Wonder 1.0 remain recorded as untracked Class B facts; no tracked lineup changed. | This page, 12 Oct 2026 |
| 11 Oct | 1 night, partial | Both official-source desks freshly covered all 22 routes; the independent close verified 63 of 66 Standard tiers. The three Perplexity Sonar tiers remain visibly carried because current Agent API pricing and migration guidance publish one Sonar basis without the exact context and distinct availability needed to preserve three tracked slots. No Standard amount, tracked model, exact Batch rule or Cache field changed. Batch closes at 22 supported, four unsupported and 40 unestablished. Cache closes at six supported, three explicitly unavailable, 57 unestablished and zero blocked across all 66 exact records and 22 provider routes, with observed evidence 45 priced, three explicitly unavailable, 18 unestablished and zero blocked. GPT-6.1 Sol, Claude Haiku 5.5, Mistral Large 4, Grok 4.7, Qwen3.8 Max and Muse Spark 1.3 remain held without qualifying successor maps; GPT-6.1 Sol Ultrafast remains an untracked Class B processing tier. | This page, 11 Oct 2026 |
| 10 Oct | 1 night, partial | Both official-source desks freshly covered all 22 routes; the independent close verified 63 of 66 Standard tiers. The three Perplexity Sonar tiers remain visibly carried because current Agent API pricing and migration guidance publish one Sonar basis without the exact context and distinct availability needed to preserve three tracked slots. No Standard amount or tracked model changed. Batch closes at 22 supported, four unsupported and 40 unestablished. Cache closes at six supported, three explicitly unavailable, 57 unestablished and zero blocked across all 66 exact records and 22 provider routes, with observed evidence 45 priced, three explicitly unavailable, 18 unestablished and zero blocked. GPT-6.1 Sol, Claude Haiku 5.5, Mistral Large 4, Grok 4.7, Qwen3.8 Max and Muse Spark 1.3 remain held without qualifying successor maps; GPT-6.1 Sol Ultrafast remains an untracked Class B processing tier. | This page, 10 Oct 2026 |
| 9 Oct | 1 night, partial | Both official-source desks freshly covered all 22 routes; the independent close verified 63 of 66 Standard tiers. The three Perplexity Sonar tiers remain visibly carried because current Agent API pricing and migration guidance publish one Sonar basis without the exact context and distinct availability needed to preserve three tracked slots. No Standard amount or tracked model changed. Batch closes at 22 supported, four unsupported and 40 unestablished. Cache closes at six supported, three explicitly unavailable, 57 unestablished and zero blocked across all 66 exact records and 22 provider routes, with observed evidence 45 priced, three explicitly unavailable, 18 unestablished and zero blocked. GPT-6.1 Sol, Claude Haiku 5.5, Mistral Large 4 and Grok 4.7 remain held without qualifying successor maps; GPT-6.1 Sol Ultrafast remains an untracked Class B processing tier. | This page, 9 Oct 2026 |
| 8 Oct | 1 night, partial | Both official-source desks freshly covered all 22 routes; the independent close verified 63 of 66 Standard tiers. The three Perplexity Sonar tiers remain visibly carried: current Agent API pricing and migration guidance reproduce the changed service basis but do not establish the exact context and explicit availability needed to preserve three tracked slots. Seven Will-authorized model swaps remain supported by fresh exact evidence; no same-model Standard price changed at the close. Batch closes at 22 supported, four unsupported and 40 unestablished. Cache closes at six supported, three explicitly unavailable, 57 unestablished and zero blocked across all 66 exact records and 22 provider routes, with observed evidence 45 priced, three explicitly unavailable, 18 unestablished and zero blocked. Claude Haiku 5.5 and Mistral Large 4 remain held without unique successor maps; GPT-6.1 Sol Ultrafast remains an untracked Class B processing tier. | This page, 8 Oct 2026 |
| 7 Oct | 1 night, partial | Both official-source desks freshly covered all 22 routes. The independent close verified 54 of 66 Standard tiers and carried twelve: Together GPT-OSS 20B and Kimi K2.6, all three Perplexity Sonar tiers, all three AWS Bedrock tiers, Fireworks GPT-OSS 20B and DeepSeek V4 Pro, Cerebras ZAI-GLM-4.7 and retired Kimi K2.5. Anthropic halved Claude Sonnet 5.5 cache reads from $0.20 to $0.10 per million tokens while leaving Standard, Batch and cache-write prices unchanged. Newly launched Claude Haiku 5.5 remains outside the tracked lineup because Anthropic published no explicit successor mapping from Haiku 4.5. Batch closes at 21 supported, five unsupported and 40 unestablished. Cache closes at six supported, three explicitly unavailable, 51 unestablished and six blocked, with observed evidence 41 priced, three explicitly unavailable, 16 unestablished and six blocked. | This page, 7 Oct 2026 |
| 6 Oct | 1 night, partial | Both official-source desks freshly covered all 22 routes. The independent close verified 54 of 66 Standard tiers and carried twelve: Together GPT-OSS 20B and Kimi K2.6, all three Perplexity Sonar tiers, all three AWS Bedrock tiers, Fireworks GPT-OSS 20B and DeepSeek V4 Pro, Cerebras ZAI-GLM-4.7 and retired Kimi K2.5. Two fresh closing reads of Fireworks' exact DeepSeek V4 Pro card said serverless is not supported, reversing the opening observation; the unchanged $1.32/$3.96 history remains grey. Mistral Large 4 remains a held Public Preview launch pending Will's decision. Batch remains 21 supported, five unsupported and 40 unestablished. Cache closes at six supported, three explicitly unavailable, 51 unestablished and six blocked across all 66 exact records and 22 provider routes, with observed evidence 41 priced, three explicitly unavailable, 16 unestablished and six blocked. | This page, 6 Oct 2026 |
| 5 Oct | 1 night, partial | Both official-source desks freshly covered all 22 routes. The independent close initially verified 54 of 66 Standard tiers and carried twelve. At the 11:10 watchdog, two fresh first-party reads reproduced the exact Fireworks DeepSeek V4 Pro (0813) card at the tracked $1.32 input and $3.96 output rate, recovering that unchanged same-model point to verified. The final record verifies 55 and carries eleven: Together GPT-OSS 20B and Kimi K2.6, all three Perplexity Sonar tiers, all three AWS Bedrock tiers, Fireworks GPT-OSS 20B, Cerebras ZAI-GLM-4.7 and retired Kimi K2.5. Batch remains 21 supported, five unsupported and 40 unestablished. Cache remains six supported, three explicitly unavailable, 52 unestablished and five blocked across all 66 exact records and 22 provider routes, with observed evidence 42 priced, three explicitly unavailable, 16 unestablished and five blocked. No tracked model, list price, Batch rule or Cache activation field changed. | This page, 5 Oct 2026 |
| 4 Oct | 1 night, partial | Both official-source desks freshly covered all 22 routes and independently closed at 54 of 66 Standard tiers verified. Twelve exact-model holds remain visibly carried: Together GPT-OSS 20B and Kimi K2.6, all three Perplexity Sonar tiers, all three AWS Bedrock tiers after rendered provider panels did not reproduce exact human-readable amounts, Fireworks GPT-OSS 20B and DeepSeek V4 Pro, Cerebras ZAI-GLM-4.7 and retired Kimi K2.5. The 8 PM pricing read reproduced all three legacy Perplexity rows; the 9:10 watchdog and independent close then omitted them while the official migration benchmark continued to map Sonar and Sonar Pro to Agent API fast and Sonar Reasoning Pro to low, so the prior rates remain grey. Batch closed at 21 supported, five unsupported and 40 unestablished. Cache closed at six supported, three explicitly unavailable, 52 unestablished and five blocked across all 66 exact records and 22 provider routes. No Standard amount, Batch rule or Cache field changed at the independent close. | This page, 4 Oct 2026 |
| 3 Oct | 1 night, missed | The October 2 newsroom task retained an ordinary work lease after it stalled, so the October 3 official desks, watchdogs and morning continuity backstop all refused to write through another agent. No October 3 source pass or closing receipt exists. On October 4 all 66 tiers were explicitly carried from the October 2 record, leaving the complete October 3 row permanently grey rather than backdating verification. | This page, 4 Oct 2026 |
| 2 Oct | 1 night, opening partial | The 8 PM official-source desk freshly covered all 22 routes and verified 56 of 66 Standard tiers. Ten exact-model holds remain visibly carried: Together GPT-OSS 20B and Kimi K2.6, all three Perplexity Sonar tiers, AWS Claude Sonnet 5, Fireworks GPT-OSS 20B and DeepSeek V4 Pro, Cerebras ZAI-GLM-4.7 and retired Kimi K2.5. Batch remains 21 supported, five unsupported and 40 unestablished. Cache remains six supported, three explicitly unavailable, 52 unestablished and five blocked across all 66 exact records and 22 provider routes. Z.ai's migration guide explicitly directed GLM-5.2 users to generally available GLM-5.3, so the same Flagship slot moved prospectively at the unchanged $1.40/$4.40 standard rate under the guarded-successor rule; historical GLM-5.2 points remain intact. The independent 10 PM close is still required. | This page, 2 Oct 2026 |
| 1 Oct | 1 night, partial | Both official-source desks freshly covered all 22 routes. The independent close initially verified 54 of 66 Standard tiers and carried 12. At the 11:10 watchdog, fresh rendered US East (Ohio) panels reproduced Claude Haiku 4.5 at $1/$5 and Llama 4 Maverick at $0.24/$0.97, restoring both to verified. The final record verifies 56 and carries ten: Together GPT-OSS 20B and Kimi K2.6, all three Perplexity Sonar tiers, AWS Claude Sonnet 5, Fireworks GPT-OSS 20B and DeepSeek V4 Pro, Cerebras ZAI-GLM-4.7 and retired Kimi K2.5. Batch remains 21 supported, five unsupported and 40 unestablished. Cache remains six supported, three explicitly unavailable, 52 unestablished and five blocked across all 66 exact records and 22 provider routes. DeepInfra's provider-directed Gemma 4 31B Turbo successor remains current at $0.09/$0.34, with incomplete secondary conditions kept unestablished. | This page, 1 Oct 2026 |
| 30 Sep | 1 night, partial | Both official-source desks freshly covered all 22 routes, and the independent close verified 56 of 66 Standard tiers. Together GPT-OSS 20B and Kimi K2.6, all three Perplexity Sonar tiers, AWS Claude Sonnet 5, Fireworks GPT-OSS 20B and DeepSeek V4 Pro, Cerebras ZAI-GLM-4.7 and retired Kimi K2.5 remain visibly carried. AWS Haiku 4.5 and Llama 4 Maverick returned to verified after their current Ohio rows reproduced. Batch remains 21 supported, five unsupported and 40 unestablished. Cache remains six supported, three explicitly unavailable, 52 unestablished and five blocked across all 66 exact records and 22 provider routes. DeepInfra still serves tracked Llama 4 Maverick at $0.20/$0.80 before its scheduled October 1 automatic redirect, so no future successor was applied early. | This page, 30 Sep 2026 |
| 29 Sep | 1 night, opening partial | The 8 PM official-source desk freshly covered all 22 routes and verified 54 of 66 Standard tiers. Together GPT-OSS 20B and Kimi K2.6, all three Perplexity Sonar tiers, all three AWS tiers, Fireworks GPT-OSS 20B and DeepSeek V4 Pro, Cerebras ZAI-GLM-4.7 and retired Kimi K2.5 remain visibly carried. OpenRouter's exact endpoint API reproduced tracked DeepSeek V3 0324 on a new cheapest SiliconFlow route at $0.25/$1, replacing the former DeepInfra $0.24/$0.90 route; the same-model price change is recorded prospectively. OpenAI launched GPT-6.1 Sol at $2/$10 with exact Batch and Cache terms, but published no successor mapping from tracked GPT-6 Sol, so no lineup change was made. Batch remains 21 supported, five unsupported and 40 unestablished. Cache remains six supported, three explicitly unavailable, 52 unestablished and five blocked across all 66 exact records and 22 provider routes; OpenRouter's observed cached-input evidence moved from $0.135 on the withdrawn DeepInfra route to $0.11 on GMICloud without changing public Cache activation. | This page, 29 Sep 2026 |
| 28 Sep | 1 night, partial | Both official-source desks freshly covered all 22 routes, and the independent close verified 56 of 66 Standard tiers. Together GPT-OSS 20B and Kimi K2.6, all three Perplexity Sonar tiers, AWS Claude Sonnet 5, Fireworks GPT-OSS 20B and DeepSeek V4 Pro, Cerebras ZAI-GLM-4.7 and retired Kimi K2.5 remain visibly carried. Anthropic's first-party migration guide qualified Claude Sonnet 5.5 as the same-provider Mid-tier successor at unchanged $2/$10 Standard, $1/$5 Batch and exact Cache terms; the guarded swap preserves Sonnet 5 history. Closing reads reproduced all three Perplexity rates but confirmed that its migration changes the service basis and collapses two rows into Agent API fast, so no automatic Perplexity swap was made. Batch closed at 21 supported, five unsupported and 40 unestablished. Cache closed at six supported, three explicitly unavailable, 52 unestablished and five blocked across all 66 exact records and 22 provider routes. No secondary-pricing rule changed. | This page, 28 Sep 2026 |
| 27 Sep | 1 night, partial | Both official-source desks covered all 22 routes and verified 60 of 66 Standard tiers. The 11:10 PM watchdog recovered the missing independent close with fresh reads across all 66 tiers. Together GPT-OSS 20B and Kimi K2.6, AWS Claude Sonnet 5, Fireworks GPT-OSS 20B, Cerebras ZAI-GLM-4.7 and retired Kimi K2.5 remain visibly carried on exact current evidence. Two fresh Together model-page reads reproduced its two rates, but the distinct current serverless catalog still omitted both, so neither point was upgraded. Batch remains 21 supported, five unsupported and 40 unestablished. Cache remains six supported, three explicitly unavailable, 52 unestablished and five blocked across all 66 exact records. No Standard, Batch or Cache fact changed at the close. | This page, 27 Sep 2026 |
| 26 Sep | 1 night, partial | Both official-source desks covered all 22 routes. The 11:10 PM watchdog recovered the missing independent close with fresh reads across all 66 Standard tiers, verifying 60 and carrying six: Together GPT-OSS 20B and Kimi K2.6, AWS Claude Sonnet 5, Fireworks GPT-OSS 20B, Cerebras ZAI-GLM-4.7 and retired Kimi K2.5. Perplexity's three token rates reproduced, while its September 27 lifecycle sentence remained absent from both rendered first-party pages and therefore stays withdrawn from current copy. Both desks reconciled all 66 exact Batch and Cache records; no secondary rule changed. | This page, 26 Sep 2026 |
| 25 Sep | 1 night, partial | Both official-source desks covered all 22 routes. The independent close initially lost Claude Haiku 4.5 from its rendered AWS table. At the 11:10 PM watchdog, two fresh cache-busted AWS Ohio renders reproduced Haiku 4.5 at $1/$5 with Batch $0.50/$2.50 and Cache read $0.10, and the distinct current AWS Batch model list named Haiku 4.5. The watchdog also corrected v1282's crossed provenance, which had marked Anthropic Haiku carried while leaving AWS Haiku verified. The final record verifies 60 of 66 Standard tiers and carries six: Together GPT-OSS 20B and Kimi K2.6, AWS Claude Sonnet 5, Fireworks GPT-OSS 20B, Cerebras ZAI-GLM-4.7 and retired Kimi K2.5. Both desks reconciled all 66 exact Batch and Cache records; no secondary rule changed. | This page, 25 Sep 2026 |
| 24 Sep | 1 night, partial | Both official-source desks covered all 22 routes. The independent close verified 60 of 66 Standard tiers and carried six: Together GPT-OSS 20B and Kimi K2.6, AWS Claude Sonnet 5, Fireworks GPT-OSS 20B, Cerebras ZAI-GLM-4.7 and retired Kimi K2.5. Perplexity's current migration page again states that Sonar support continues through September 27. DeepInfra's main table and OpenRouter's model list omitted one tracked row, but fresh exact-model pages reproduced DeepInfra Llama 4 Maverick at $0.20/$0.80 and OpenRouter DeepSeek V3 0324 at $0.24/$0.90 with $0.135 cached input. Both desks reconciled all 66 exact Batch and Cache records; no secondary rule changed. | This page, 24 Sep 2026 |
| 23 Sep | 1 night, partial | Both official-source desks covered all 22 routes and verified 60 of 66 Standard tiers. The independent close reproduced Claude Haiku 4.5 at $1/$5 and Llama 4 Maverick at $0.24/$0.97 in the rendered US East (Ohio) tables. Claude Sonnet 5 remained absent while nearby Sonnet 4.6 and 4.5 rows rendered, so its prior $2/$10 rate stays visibly carried with Together GPT-OSS 20B and Kimi K2.6, Fireworks GPT-OSS 20B, Cerebras ZAI-GLM-4.7 and retired Kimi K2.5. All 66 exact Batch and Cache records were reconciled without changing a secondary rule. Anthropic's new Claude Opus 5.5 launch was recorded separately without changing the tracked lineup. | This page, 23 Sep 2026 |
| 22 Sep | 1 night, partial | Both official-source desks covered all 22 routes and verified 60 of 66 Standard tiers. The independent close again found that AWS's rendered Anthropic table reproduced Claude Haiku 4.5 at $1/$5 but omitted tracked Claude Sonnet 5 while listing Sonnet 4.6 and 4.5; the Meta table reproduced Llama 4 Maverick at $0.24/$0.97. Sonnet 5's prior $2/$10 rate therefore stayed visibly carried. Together GPT-OSS 20B and Kimi K2.6, Fireworks GPT-OSS 20B, Cerebras ZAI-GLM-4.7 and retired Kimi K2.5 remained the other exact-model holds. All 66 exact Batch and Cache records were reconciled in both desks; no secondary rule changed. | This page, 22 Sep 2026 |
| 21 Sep | 1 night, partial | Both official-source desks covered all 22 routes. The independent close reproduced AWS Claude Haiku 4.5 at $1/$5, Claude Sonnet 5 at $2/$10 and Llama 4 Maverick at $0.24/$0.97 in the rendered US East (Ohio) tables, resolving the opening source gap without changing a rate. The final record verifies 61 of 66 Standard tiers and carries five: Together GPT-OSS 20B and Kimi K2.6, Fireworks GPT-OSS 20B, Cerebras ZAI-GLM-4.7 and retired Kimi K2.5. xAI launched generally available Grok 4.7 at $2/$6 below 200K context with $0.50 cached input, but did not explicitly replace tracked Grok 4.5, so no lineup changed. All 66 exact Batch and Cache records were reconciled; no secondary rule changed. | This page, 21 Sep 2026 |
| 20 Sep | 1 night, partial | Both official-source desks covered all 22 routes. The independent close verified 61 of 66 Standard tiers and carried five: Together GPT-OSS 20B and Kimi K2.6, Fireworks GPT-OSS 20B, Cerebras ZAI-GLM-4.7 and retired Kimi K2.5. OpenRouter DeepSeek V3 0324 reproduced on one current DeepInfra route at $0.24/$0.90 with $0.135 cached input. At the 11:10 PM watchdog, Perplexity's canonical pricing page, current Sonar catalog and exact model pages all omitted the September 27 support sentence while retaining the three unchanged token rates, so TokenScale withdrew only that lifecycle claim. Both desks reconciled all 66 exact Batch and Cache records; no secondary rule changed. | This page, 20 Sep 2026 |
| 19 Sep | 1 night, partial | Both official-source desks covered all 22 routes. The independent close verified 61 of 66 Standard tiers and carried five: Together GPT-OSS 20B and Kimi K2.6, Fireworks GPT-OSS 20B, Cerebras ZAI-GLM-4.7 and retired Kimi K2.5. OpenRouter DeepSeek V3 0324 reproduced on one current DeepInfra route at $0.24/$0.90 with $0.135 cached input. Perplexity still published all three Sonar rates, but its closing page no longer published the September 27 support sentence seen at 8 PM, so current lifecycle copy was withdrawn. Both desks reconciled all 66 exact Batch and Cache records; no secondary rule changed. | This page, 19 Sep 2026 |
| 18 Sep | 1 night, partial | Both official-source desks covered all 22 routes. The opening desk initially verified 60 of 66 Standard tiers. At the 9:10 PM watchdog, two fresh OpenRouter exact-endpoint reads reproduced one DeepInfra DeepSeek V3 0324 route at the unchanged $0.24/$0.90 pair and $0.135 cached input, while a distinct current model-list API confirmed the exact model. The independent close reproduced that route again and restored Perplexity's September 27 Sonar support sentence from its controlling first-party page. The final record is 61 verified and five carried: Together GPT-OSS 20B and Kimi K2.6, Fireworks GPT-OSS 20B, Cerebras ZAI-GLM-4.7 and retired Kimi K2.5. Both desks reconciled all 66 exact Batch and Cache records; public Cache activation coverage stayed unchanged. | This page, 18 Sep 2026 |
| 17 Sep | 1 night, opening partial | The 8 PM official-source desk covered all 22 routes and verified 61 of 66 Standard tiers. Together GPT-OSS 20B and Kimi K2.6, Fireworks GPT-OSS 20B, Cerebras ZAI-GLM-4.7 and retired Kimi K2.5 remained exact-model holds, so their prior rates stayed visibly carried. All 66 exact Batch and Cache records were reconciled. Perplexity's three rates remained unchanged, but its current page no longer published the earlier September 27 support sentence, so current lifecycle copy was withdrawn pending the independent 10 PM close. | This page, 17 Sep 2026 |
| 16 Sep | 1 night, partial | Both official-source desks covered all 22 routes and closed at 61 of 66 Standard tiers verified. Together GPT-OSS 20B and Kimi K2.6, Fireworks GPT-OSS 20B, Cerebras ZAI-GLM-4.7 and retired Kimi K2.5 remained exact-model holds, so their prior rates stayed visibly carried. All 66 exact Batch and Cache records were reconciled in both passes. Perplexity's three rates remained unchanged while its independent close again stated that Sonar support continues through September 27, confirming the restored lifecycle fact. | This page, 16 Sep 2026 |
| 15 Sep | 1 night, partial | Both official-source desks covered all 22 routes. The opening desk initially verified 60 of 66 Standard tiers. At the 9:10 PM watchdog, two fresh cache-busted OpenRouter endpoint reads reproduced one DeepInfra DeepSeek V3 0324 route at the unchanged $0.24/$0.90 pair, with a distinct current model-list API confirming the exact model. The independent close reproduced that route again, leaving the final record at 61 verified and five carried: Together GPT-OSS 20B and Kimi K2.6, Fireworks GPT-OSS 20B, Cerebras ZAI-GLM-4.7 and retired Kimi K2.5. All 66 exact Batch and Cache records were reconciled in both scheduled branches. | This page, 15 Sep 2026 |
| 14 Sep | 1 night, partial | Both official-source desks covered all 22 routes and closed at 60 of 66 Standard tiers verified. Together GPT-OSS 20B and Kimi K2.6 remained absent from current serverless inference pricing, and similarly named fine-tuning rows did not establish inference rates. Fireworks GPT-OSS 20B was absent from pricing and its exact page said Serverless: Not supported; Cerebras ZAI-GLM-4.7 was absent from the complete current catalog; Kimi K2.5 was retired, and its directed K3 successor failed the same-tier guard; and OpenRouter DeepSeek V3 had split input and output across separate endpoints. Those six prior rates stayed visibly carried. All 66 exact Batch records and all 66 exact Cache records across 22 provider routes were reconciled. | This page, 17 Sep 2026 |
| 13 Sep | 1 night, partial | Both official-source desks covered all 22 routes and closed at 60 of 66 Standard tiers verified. Together GPT-OSS 20B and Kimi K2.6 had no exact current serverless inference rows; Fireworks GPT-OSS 20B had no rate and its exact page said Serverless: Not supported; Cerebras ZAI-GLM-4.7 was absent from the complete catalog and its exact endpoint returned 404; retired Kimi K2.5 had no unique provider-directed successor; and OpenRouter DeepSeek V3 split the lowest input and output across different endpoints. Those six prior rates stayed visibly carried rather than mixing routes or changing the tracked lineup. All 66 exact Batch records and all 66 exact Cache records across 22 provider routes were reconciled. | This page, 17 Sep 2026 |
| 12 Sep | 1 night, partial | Both official-source desks covered all 22 routes and verified 60 of 66 Standard tiers. Together GPT-OSS 20B and Kimi K2.6, Fireworks GPT-OSS 20B, Cerebras ZAI-GLM-4.7, retired Kimi K2.5 and OpenRouter's split DeepSeek V3 route remained exact-model holds, so their prior rates stayed visibly carried. All 66 exact Batch and Cache records were reconciled. Current Cohere model pages and the catalog reproduced all three rates and Live status after the landing-page table disappeared. | This page, 16 Sep 2026 |
| 11 Sep | 1 night, partial | Both official-source desks opened all 22 routes and verified 60 of 66 Standard tiers. Together GPT-OSS 20B and Kimi K2.6, Fireworks GPT-OSS 20B, Cerebras ZAI-GLM-4.7, retired Kimi K2.5 and OpenRouter's split DeepSeek V3 route remained exact-model holds, so their prior rates stayed visibly carried. The opening desk also recorded that OpenRouter's former single V3 route and cached-input field had disappeared; the independent close reproduced the split routes. All 66 exact Batch and Cache records were reconciled. | This page, 15 Sep 2026 |
| 10 Sep | 1 night, partial | Both official-source desks opened all 22 routes and the closing watchdog recovered MiniMax M2 from its current API overview and explicit-cache guide, leaving 61 of 66 Standard tiers verified. Together GPT-OSS 20B and Kimi K2.6, Fireworks GPT-OSS 20B, Cerebras ZAI-GLM-4.7 and retired Kimi K2.5 remained exact-model holds, so their prior rates stayed visibly carried. DeepSeek Lite moved prospectively to the provider-directed V4.1 Flash successor after the complete guarded-successor review passed. All 66 exact Batch and Cache entries were reconciled. | This page, 14 Sep 2026 |
| 9 Sep | 1 night, partial | The 8 PM official-source desk opened all 22 routes and verified 60 of 66 Standard tiers. Together GPT-OSS 20B and Kimi K2.6, Fireworks GPT-OSS 20B, Cerebras ZAI-GLM-4.7, retired Kimi K2.5 and legacy MiniMax M2 remained exact-model holds, so their prior rates stayed visibly carried. All 66 exact Batch and Cache entries were reconciled. Perplexity's three rates verified while its earlier September 27 support sentence was absent and withdrawn from current copy. | This page, 13 Sep 2026 |
| 8 Sep | 1 night, partial | The two official-source desks opened all 22 routes, and the closing watchdog resolved all three rendered AWS Bedrock holds without changing their rates. The final record verifies 60 of 66 Standard tiers. Together GPT-OSS 20B and Kimi K2.6, Fireworks GPT-OSS 20B, Cerebras ZAI-GLM-4.7, retired Kimi K2.5 and legacy MiniMax M2 remained exact-model holds, so their prior rates stayed visibly carried. All 66 exact Batch and Cache entries were reconciled. | This page, 12 Sep 2026 |
| 7 Sep | 1 night, partial | The recovered closing desk freshly opened all 22 official routes and verified 57 of 66 Standard tiers. Together Small and Flagship, AWS Bedrock's three rendered rows, Fireworks Small, Cerebras Flagship, retired Kimi K2.5 and legacy MiniMax M2 remained exact-model holds. All nine prior rates stayed visibly carried, and all 66 exact Batch and Cache entries were reconciled without inventing replacements. | This page, 11 Sep 2026 |
| 6 Sep | 1 night, partial | Both official-source desks freshly opened all 22 provider routes and verified 60 of 66 Standard tiers. Together GPT-OSS 20B and Kimi K2.6, Fireworks GPT-OSS 20B, Cerebras ZAI-GLM-4.7, retired Kimi K2.5 and legacy MiniMax M2 remained exact-model holds, so their prior rates stayed visibly carried rather than being silently replaced. The closing desk also reconciled all 66 exact Batch and Cache entries. | This page, 10 Sep 2026 |
| 5 Sep | 1 night, partial | The 8 PM desk freshly opened all 22 official routes and initially verified 59 of 66 standard tiers. At 9:10, two fresh cache-busted Cohere pricing reads reproduced Command R+ (08-2024) at the unchanged $2.50/$10 rate, while Cohere's separate current model table still marked the exact SKU Live. The same-day point was upgraded, leaving 60 verified and 6 carried exact-model holds: Together GPT-OSS 20B and Kimi K2.6, Fireworks GPT-OSS 20B, Cerebras ZAI-GLM-4.7, retired Kimi K2.5 and legacy MiniMax M2. All 66 Batch entries and all 22 Cache provider routes were compared. | This page, 5 Sep 2026 |
| 4 Sep | 1 night, missed in shipping history | The separate September 4 desks and durable ledger preserve an original closing result of 60 verified and 6 carried tiers. A later local feature chain was mistakenly rebased on the live September 3 build and dropped all 66 September 4 points from the shipping data. The September 5 recovery restored the missing public date conservatively as 66 carried points instead of overstating what the replacement candidate itself could prove. | This page, 5 Sep 2026 |
| 3 Sep | 1 night, partial | The 8 PM desk freshly opened all 22 official routes and verified 61 of 66 standard tiers. Together GPT-OSS 20B and Kimi K2.6, Cerebras ZAI-GLM-4.7 and Fireworks GPT-OSS 20B remained absent or unsupported, while Moonshot's Kimi K2.5 remained explicitly retired with several possible successors. Those five rates were carried rather than silently changing the lineup. Anthropic's explicit generally available Fable 5.1 successor passed the guarded model-change gate at the same standard and Batch rates, with its exact Cache terms recorded. All 66 Batch entries and the 22-route Cache watch were also compared. | This page, 3 Sep 2026 |
| 2 Sep | 1 night, partial | Both desks freshly opened all 22 official routes. The 10 PM close verified 61 of 66 standard tiers. Together GPT-OSS 20B was reproducible at 8 PM but disappeared from both current first-party Together surfaces at close; Together Kimi K2.6, Cerebras ZAI-GLM-4.7 and Fireworks GPT-OSS 20B remained absent, while Moonshot's Kimi K2.5 remained explicitly retired with several possible successors. The previous same-model rates were carried rather than silently changing the lineup. All 66 Batch entries and the 22-route Cache watch were also compared in both branches. | This page, 2 Sep 2026 |
| 1 Sep | 1 night, missed | The scheduled 8 PM and 10 PM desks and both watchdogs were blocked while candidate v1053 awaited review and live remained v1052. No official-source route was opened for the September 1 record. Approval arrived after midnight, when a new read could no longer prove the prior date, so all 66 tiers were explicitly carried from 31 August and remain grey. No closing-desk or secondary-ledger evidence is claimed for the missed night. | This page, 2 Sep 2026 |
| 25–31 Aug | 7 nights, ended 1 Sep | Together’s tracked Kimi K2.6 and Cerebras’s tracked ZAI-GLM-4.7 remained absent from their current public catalogs, so the 25 and 26 August records verified 64 of 66 tiers and carried those two flagship rates grey. On 27 and 28 August, OpenRouter’s DeepSeek V3 route also split its lowest input and output across different endpoints, so those records verified 63 of 66 rather than inventing a mixed route. One exact DeepInfra route again owned the full $0.24/$0.90 pair on 29 and 30 August, returning OpenRouter Mid to verified. The 30 August first desk verified 63 of 66: Together and Cerebras remained grey, and Fireworks GPT-OSS 20B joined them after its rendered model page said Serverless: Not supported while the static serverless pricing table exposed no current row. The prior $0.07/$0.30 rate stayed grey and no successor was chosen silently. On 31 August the record verified 62 of 66; Together, Cerebras and Fireworks remained carried, and Moonshot’s retired Kimi K2.5 route was also carried rather than silently replaced. | This page, 29 Aug 2026 · extended 2 Sep 2026 |
| 18–24 Aug | 7 nights, ended 25 Aug | Cerebras’s tracked ZAI-GLM-4.7 remained absent from its complete public model list, so the flagship rate held grey at $2.25/$2.75 on all seven nights. The 18 to 20 August records each verified 65 of 66 tiers. On 21 August, Together’s rendered inference table also stopped listing tracked Kimi K2.6, so its $1.20/$4.50 flagship rate was carried and the record closed at 64 of 66. Together's row returned at the 22 August 8 PM read but disappeared again at the independent 10 PM close. The recovered 23 and 24 August full passes again found the row absent, so both nights verify 64 of 66 and carry the same two flagship rates grey. No replacement was chosen silently. | This page, 19 Aug 2026 · extended 24 Aug 2026 |
| 16–17 Aug | 2 nights, ended 18 Aug | DeepSeek replaced flat V4 pricing with separate peak and off-peak rates on 16 August. All three DeepSeek tiers were held rather than have the record pick a time window on the reader’s behalf, so that night closed at 63 of 66 verified. On 17 August three more first-party routes stopped reading at once: Groq’s small and mid rows during its production-model change, MiniMax’s rendered page hiding the tracked legacy row, and Cerebras’s public model list no longer containing the tracked ZAI-GLM-4.7. That night recorded 59 of 66, the lowest since 10 August. The run ended on 18 August, when the peak rate was adopted as the conservative canonical basis and the Groq lineup was settled, and both providers verified again. | This page, 19 Aug 2026 |
| 14–15 Aug | 2 nights, ended 16 Aug | The 8 PM official-source desk initially verified 64 of 66 tiers. At the independent 10 PM close, Kimi K2.5's direct official USD model page reproduced $0.60/$3, upgrading that same-day point without adding a duplicate. The final record is 65 verified and one carried: OpenRouter's previous DeepSeek V3 route disappeared, while current routes split the lowest input and output prices between two providers. That same route was still unreadable on 15 August, so the tier carried a second night and the board again recorded 65 of 66 verified. The 16 August close ended the run: fresh endpoint reads found one reproducible cheapest route for each pinned tier and that night verified at the unchanged $0.24/$0.90. | This page, 14 Aug 2026 · 15 Aug added here, 19 Aug 2026 |
| 10 Aug | 1 night | The official-console pass verified 18 of 66 tiers across Cohere, SambaNova, Cerebras, OpenRouter, xAI and Meta. The other 48 points were explicitly carried because their official routes were not completed that day. No carried point was presented as freshly checked. | This page, 11 Aug 2026 |
| 6–9 Aug | 4 nights, ended 11 Aug | A second run with no full verification pass, days after the first. Every tier was carried forward on 6 and 7 August. The 8 August run reached only 3 of the 66 tracked tiers before it stopped. Daytime passes on 9 August eventually read 48 of 66 clean, with zero price changes, while 18 tiers remained behind missing or rendered-only routes or model-choice decisions. The run ended on 11 August when all 66 tiers received a same-day official-source check. | This page, 9 Aug 2026 · closed 11 Aug 2026 |
| 1–4 Aug | 4 nights, ended 5 Aug | The nightly kept appending data but no verification pass ran and no provider was stamped as checked. Found on 4 August while auditing the site against its own record. The run went one night longer than this page first disclosed: 4 August was carried forward in full as well. It ended on 5 August, when a pass verified 39 of the 66 tracked tiers and held the other 27. | This page + journal, 4 Aug 2026 · extended and closed here, 9 Aug 2026 |
| 13–17 Jul | 5 nights | A macOS update reset a permission the scheduled job needed, and it failed quietly five nights in a row. Not detected from inside the system; it took three nights to notice. The job now raises an alarm the moment it fails. | Changelog, 18 Jul 2026 |
| 13 Jun | 1 night | The nightly carried every tier forward unchanged without reading the sources. A full re-verification the next day confirmed most values as correct and fixed six tier mappings it found along the way. | This page, 4 Aug 2026 |
| 31 May | 1 night | Both the primary and the backup nightly jobs skipped the run entirely. The miss was found the next day and the previous prices were held. | This page, 4 Aug 2026 |
| 27–30 May | 4 nights | The nightly job wrote prices it could not actually confirm at source. A data audit on 31 May caught it and reverted all 36 affected points to the last verified values. The public site never showed the bad numbers, but those nights were not verified. | This page, 4 Aug 2026 |
Partly verified nights. 93 nights read part of the board and held the rest: 5 August (39 of 66 tiers verified), 9 August (45 of 66), 10 August (18 of 66), 14 August (65 of 66), 15 August (65 of 66), 16 August (63 of 66), 17 August (59 of 66), 18 August (65 of 66), 19 August (65 of 66), 20 August (65 of 66), 21 August (64 of 66), 22 August (64 of 66), 23 August (64 of 66), 24 August (64 of 66), 25 August (64 of 66), 26 August (64 of 66), 27 August (63 of 66), 28 August (63 of 66), 29 August (64 of 66), 30 August (63 of 66), 31 August (62 of 66), 2 September (61 of 66), 3 September (61 of 66), 4 September (60 of 66), 5 September (60 of 66), 6 September (60 of 66), 7 September (57 of 66), 8 September (60 of 66), 9 September (60 of 66), 10 September (61 of 66), 11 September (60 of 66), 12 September (60 of 66), 13 September (60 of 66), 14 September (60 of 66), 15 September (61 of 66), 16 September (61 of 66), 17 September (61 of 66), 18 September (61 of 66), 19 September (61 of 66), 20 September (61 of 66), 21 September (61 of 66), 22 September (60 of 66), 23 September (60 of 66), 24 September (60 of 66), 25 September (60 of 66), 26 September (60 of 66), 27 September (60 of 66), 28 September (56 of 66), 29 September (54 of 66), 30 September (56 of 66), 1 October (56 of 66), 2 October (56 of 66), 4 October (54 of 66), 5 October (55 of 66), 6 October (54 of 66), 7 October (54 of 66), 8 October (63 of 66), 9 October (63 of 66), 10 October (63 of 66), 11 October (63 of 66), 12 October (63 of 66), 13 October (63 of 66), 14 October (63 of 66), 15 October (63 of 66), 16 October (63 of 66), 17 October (63 of 66), 18 October (62 of 66), 19 October (61 of 66), 20 October (58 of 66), 21 October (58 of 66), 22 October (58 of 66), 23 October (56 of 66), 24 October (56 of 66), 25 October (56 of 66), 26 October (56 of 66), 27 October (56 of 66), 28 October (56 of 66), 29 October (56 of 66), 30 October (56 of 66), 31 October (59 of 66), 1 November (59 of 66), 2 November (59 of 66), 3 November (59 of 66), 4 November (59 of 66), 6 November (59 of 66), 7 November (56 of 66), 8 November (56 of 66), 10 November (56 of 66), 14 November (56 of 66), 15 November (56 of 66), 16 November (56 of 66), 17 November (56 of 66) and 18 November (56 of 66). Each is covered by its row in the table above and counted as a miss for every tier it did not reach. Those tiers stay grey on their charts. These counts are the points the record actually holds for each night; a row above may describe a different number of tiers as read, because reading a page and writing down a verified price are not the same act.
Correction, 9 August 2026. This page understated its own record. From 4 to 9 August it showed 14 carried-forward nights while the price record held 17, left the August run marked ongoing after it had ended, and did not disclose the nights of 4, 6 and 7 August at all. On the page whose whole job is disclosure, that is the worst place to be wrong, so it is stated here rather than quietly fixed. Two things changed with it: the counts above are now computed from the price record itself on every build, and the build now fails if any night in the record is not covered by a row in this table. Logged in the changelog.
How this table was built, honestly. Before 4 August 2026 the record did not store each night's verified-or-carried status; it can now, permanently, in the data itself. The rows above are reconstructed from build logs and the change log. If a missed night ever surfaces that this table does not show, it gets added, dated, with the reason, and the correction is noted in the changelog.
Verification is per provider page, not all-or-nothing. On most verified nights all 22 pages read cleanly; on some, a page that will not render without scripts gets held while the rest verify. A rendered-source pass on 11 August cleared the current holds. Cohere is pinned to the live dated Command R7B 12-2024, Command R 08-2024 and Command R+ 08-2024 rows. AWS Bedrock is pinned to Claude Haiku 4.5, Claude Sonnet 5 and active Llama 4 Maverick 17B Instruct, with its exact US East (Ohio) rate. Kimi K2.5 was also matched on its rendered official model page. Earlier carried dates remain grey permanently; the successful 11 August check does not rewrite those gaps.
List price is not effective price. Prices are recorded as published, in dollars per million tokens. On 30 June 2026, Claude Sonnet 5 shipped with a new tokenizer that turns the same text into roughly 30% more tokens at an unchanged per-token rate: a real cost increase that never appeared on a pricing page. TokenScale records the published rate; it does not yet measure cost per unit of actual work.
Token estimates are an average. Roughly 1.33 tokens per word, which holds for typical English prose and will not hold for your code, your JSON, or your Mandarin.
The nightly record starts on 19 May 2026. Anything earlier is a baseline drawn from the sources available at the time, and is marked as such on the charts.
Promotional rates never enter the history. They are shown live, clearly labelled, and excluded from the record, because a limited-time rate is not what the thing cost.
You don't have to take my word for any of it.
Starting 5 August 2026, TokenScale's key pages are submitted to the Internet Archive every night by a watchdog that runs outside the machine the nightly runs on. That builds a third-party, timestamped copy of what this site said on any given date, one I don't control and can't edit. Earlier snapshots exist too: the Archive holds captures of this site going back to its launch window, through 9 June 2026, which corroborate the early record.
The providers' own pricing pages are archived on a rotating schedule as well. Where those captures work, you can compare what TokenScale recorded against an independent snapshot of the source. Where a provider's page is JavaScript-rendered and archives as an empty shell, that provider will be marked here as not independently checkable, because pretending otherwise would defeat the point. A provider-by-provider table appears here once the watchdog has a full week of archive results to report.
Tell me and I'll correct it in public, in the changelog, within a day. Use the "Report a pricing error" link in the calculator's footer, or email projectsbilton@gmail.com. That is the whole arrangement.
Last updated: 18 November 2026. The counts above are computed from the price record on every build and run through the night of 18 Nov.