Background
Sound
TokenScale Top Text Models

Bilton Projects Presentation

2026
TokenScale

The price chart for tokens

The Billboard Chart for AI Models

📻 This week's AI models, ranked like the pop charts. Lean 💸 cheap, 🧠 brainy, ⚡ fast or 🌱 efficient below, then scrub the time machine through every date a tracked price actually changed (logged nightly since 12 May 2026) and watch the crown change hands.

📻 Pick your lean: the chart re-ranks live

Fine-tune the mix: slide to weight each side

🔊 Fader sound

📈 Current editionRates through 21 Aug 2026 LIVE
1
Gemini Flash-Lite 👑 #1 this week
Google
$0.74
per 1M tok
2
GPT-5.6 Luna
OpenAI
$0.40
per 1M tok
3
Gemini Flash
Google
$2.70
per 1M tok
4
MiniMax M3
MiniMax
$0.48
per 1M tok
5
Qwen3.7 Max
Alibaba
$3.50
per 1M tok
6
GLM-5.2
Zhipu
$2.00
per 1M tok
7
Muse Spark 1.2
Meta
$1.85
per 1M tok
8
DeepSeek V4 Pro
DeepSeek
$1.85
per 1M tok
9
GPT-5.6 Terra
OpenAI
$4.00
per 1M tok
10
Grok 4.5
xAI
$2.80
per 1M tok
11
Claude Haiku 4.5
Anthropic
$1.80
per 1M tok
12
Claude Sonnet 5
Anthropic
$3.60
per 1M tok
13
Gemini 3.1 Pro
Google
$4.00
per 1M tok
14
Kimi K3
Moonshot
$5.40
per 1M tok
15
Mistral Large 3
Mistral
$0.70
per 1M tok
16
GPT-5.6 Sol
OpenAI
$10.00
per 1M tok
17
Claude Fable 5
Anthropic
$18.00
per 1M tok
▲▼ vs. previous editionFull table & sources ↗ August 21, 2026

Time machine

Rewind the chart: who was #1 before?
AUG 21 2026
👑 #1 on this date Gemini Flash-Lite
12 May ’2620 MayJun ’26
⏱ The dial stops on the dates a tracked rate actually changed, not on every night. A quiet fortnight is one stop, not fourteen. Want one provider's night-by-night line? That lives on TokenScale's Price history →
Rec 17 models on the chart at this date
Full table & sources · TokenScale ↗
The small print

The dial stops where something actually happened in the log, not every day. Every rate on the Text chart comes from our own nightly price checks, recorded since 12 May 2026 (the same dataset behind TokenScale). Coverage grew as logging expanded: OpenAI from 12 May, Anthropic from 14 May, most providers from 15 May, Mistral from 19 May, and Qwen, GLM, Kimi, MiniMax and Llama from 2 June. Models join the chart when their logs begin, marked NEW. We used to show a hand-reconstructed 2023–2025 era here; we pulled it until it can be properly audited, because this page should only say things we can back with data. The Creative chart is still hand-curated for now. Spotted a wrong rate or date? Tell us and the history gets better.

👑
Your #1 pick
Gemini Flash-Lite · Google
Best all-round on a balanced weighting.
$0.74per 1M tok
The AI chart · ranked your way

Which model is #1? Depends what you're into.

There isn't one best AI model. The winner depends on what you're measuring. Lean toward price, brains, speed or efficiency, and watch the models climb, fall and trade the crown. It's a singles chart for AI: the ranking moves every week, and the countdown is yours. See the full price reference ↗

Sources & method: what's measured, what's estimated