The state of AI pricing, measured.
Every list-price movement across 22 providers and 66 tracked tiers, counted from the nightly record. No estimates and no forecasts: this page recomputes from the same append-only history as the rest of the site, and refreshes with every nightly build.
The last 30 days
22 movements in the window 2026-07-23 to 2026-08-21, across 66 tracked tiers.
| Date | Provider | Tier | Was (in / out) | Now (in / out) | What |
|---|---|---|---|---|---|
| 2026-08-18 | DeepSeek | V4 Flash · Lite | $0.14 / $0.28 | $0.44 / $1.32 | raise in +214%, out +371% |
| 2026-08-18 | DeepSeek | V4 Pro · Mid | $0.435 / $0.87 | $1.32 / $3.96 | raise in +203%, out +355% |
| 2026-08-18 | DeepSeek | V4 Pro · Pro | $0.435 / $0.87 | $1.32 / $3.96 | raise in +203%, out +355% |
| 2026-08-18 | Groq | GPT-OSS 20B · Lite | $0.05 / $0.08 | $0.075 / $0.3 | model swap |
| 2026-08-18 | Groq | GPT-OSS 120B · Mid | $0.59 / $0.79 | $0.15 / $0.6 | model swap |
| 2026-08-12 | OpenRouter | DeepSeek V3 · Mid | $0.2574 / $1.03 | $0.24 / $0.9 | cut in −7%, out −13% |
| 2026-08-11 | AWS Bedrock | Llama 4 Maverick 17B (Bedrock) · Pro | $2.40 / $2.40 | $0.24 / $0.97 | model swap |
| 2026-08-10 | OpenRouter | DeepSeek V3 · Mid | $0.2002 / $0.8001 | $0.2574 / $1.03 | raise in +29%, out +29% |
| 2026-08-10 | xAI Grok | Grok 4.5 · Pro | $1.25 / $2.50 | $2.00 / $6.00 | model swap |
| 2026-08-10 | Meta | Muse Spark 1.2 · Lite | $0.11 / $0.34 | $1.25 / $4.25 | model swap |
| 2026-08-10 | Meta | Muse Spark 1.2 · Mid | $0.15 / $0.6 | $1.25 / $4.25 | model swap |
| 2026-08-10 | Meta | Muse Spark 1.2 · Pro | $0.15 / $0.6 | $1.25 / $4.25 | model swap |
| 2026-07-31 | DeepInfra | Llama 3.1 8B · Lite | $0.02 / $0.03 | $0.02 / $0.04 | raise out +33% |
| 2026-07-31 | OpenRouter | DeepSeek V3 · Mid | $0.2288 / $0.9144 | $0.2002 / $0.8001 | cut in −13%, out −12% |
| 2026-07-30 verified 2026-08-12 | OpenAI | GPT-5.6 Luna · Lite | $1.00 / $6.00 | $0.2 / $1.20 | cut in −80%, out −80% |
| 2026-07-30 verified 2026-08-12 | OpenAI | GPT-5.6 Terra · Mid | $2.50 / $15.00 | $2.00 / $12.00 | cut in −20%, out −20% |
| 2026-07-28 | Google Gemini | Flash-Lite 3.5 · Lite | $0.1 / $0.4 | $0.3 / $2.50 | model swap |
| 2026-07-28 | Google Gemini | Flash 3.6 · Mid | $1.50 / $9.00 | $1.50 / $7.50 | model swap |
| 2026-07-28 | OpenAI | GPT-5.6 Luna · Lite | $0.75 / $4.50 | $1.00 / $6.00 | model swap |
| 2026-07-28 | Groq | GPT-OSS 120B · Pro | $1.00 / $3.00 | $0.15 / $0.6 | model swap |
| 2026-07-28 | Alibaba Qwen | Qwen3.6-Flash · Lite | $0.165 / $0.99 | $0.25 / $1.50 | raise in +52%, out +52% |
| 2026-07-28 | Moonshot Kimi | Kimi K3 · Pro | $1.90 / $8.00 | $3.00 / $15.00 | model swap |
Scroll the table sideways to see every column.
A movement that coincides with a recorded model change in that slot is classed as a model swap, not a repricing: the rate moved because the model behind the tier changed. The tier column shows the current occupant of the slot. The dated story behind each move is on Price Moves; retired models rest in the Model Graveyard. A date with a quieter verified line beneath it is dated by when the price actually changed at the provider; TokenScale's nightly check confirmed it later, on the verified date.
Where the market sits
The median tracked rate rose in every tier class over the last 30 days: Lite input +51% and output +70%; Mid input +32% and output +104%; Pro input +3% and output +13%.
| Tier class | Median now (in / out) | 30 days ago (in / out) |
|---|---|---|
| Lite · 22 providers | $0.275 / $1.15 | $0.1825 / $0.675 |
| Mid · 22 providers | $1.02 / $2.85 | $0.775 / $1.40 |
| Pro · 22 providers | $1.87 / $4.45 | $1.82 / $3.94 |
Scroll the table sideways to see every column.
Median of list prices across providers tracked on both dates. Medians move only when enough of the market moves: that is the point of quoting them.
The cheapest tracked rate is Llama 3.1 8B on DeepInfra at $0.02 per 1M input tokens. The priciest tracked flagship is Claude Fable 5 at $10.00: a spread of about 500× for a million tokens of input. Capability differs as widely as price; the full ranking lists every provider.
In content terms: the cheapest tracked way to generate The Hobbit (95,356 words ≈ 127,141 tokens) as output is $0.0051 on Llama 3.1 8B (DeepInfra).
The ledger, month by month
| Month | Movements | Cuts | Raises | Model swaps |
|---|---|---|---|---|
| May 2026 | 22 | 9 | 9 | 4 |
| June 2026 | 15 | 5 | 9 | 1 |
| July 2026 | 38 | 6 | 5 | 27 |
| August 2026 | 12 | 1 | 4 | 7 |
Scroll the table sideways to see every column.
Nightly tracking began mid-May 2026, so May is a partial month. A movement dated in May can also be the first verified reading catching up with an earlier change. Historical rows are never rewritten.
About these numbers. Prices are checked nightly against each provider’s official pricing page and recorded in an append-only history (data through 2026-08-21). List prices only: promotions never enter the record. A cut lowers input and output (or one of them, with the other flat); a raise is the reverse; a movement that coincides with a recorded model change is a swap. Some movements are published corrections of the record’s own earlier reading; each is labelled as such on Price Moves. Corrections are published, not buried: see the methodology and the reliability record.