Cost and output
Every output with its grade, what the run cost, the tokens it used and how long the model worked.
Cost by test
What each model’s run of a test cost, cheapest first. Bars compare the runs of one test: every chart has its own scale.
- Gemini 4 Argon$0.67
- GPT-6.1 Sol$1.18
- Claude Haiku 5.5$1.87
- Gemini 4 Carbon$2.22
- Gemini 4 Hydrogen$2.38
- Gemini 4 Neon$5.19
- GPT-6.2 Astra$5.21
- GPT-6.1 Terra$8.17
- GPT-6.2 Sol$9.76
- GPT-6 Astra$28.51
- DeepSeek V4.1 FlashCost unknown
- Gemini 4 Argon$1.09
- GPT-6.1 Sol$1.78
- Gemini 4 Carbon$1.79
- Gemini 4 Hydrogen$2.52
- Gemini 4 Neon$4.05
- GPT-6.1 Terra$5.32
- GPT-6.2 Astra$7.41
- GPT-6.2 Sol$9.15
- GPT-6 Astra$31.82
- DeepSeek V4.1 FlashCost unknown
- Gemini 4 Argon$1.01
- GPT-6.1 Terra$1.15
- Gemini 4 Hydrogen$1.16
- GPT-6.1 Sol$1.24
- Gemini 4 Carbon$1.27
- Gemini 4 Neon$2.22
- GPT-6.2 Sol$4.01
- GPT-6.2 Astra$4.09
- Claude Haiku 5.5$12.67
- GPT-6 Astra$16.29
- Claude Opus 5.5$115.96
- Claude Sonnet 5.5$225.99
- Claude Fable 5.2$237.18
- DeepSeek V4.1 FlashCost unknown
- Gemini 4 Argon$0.95
- Gemini 4 Hydrogen$1.63
- Gemini 4 Carbon$4.01
- Gemini 4 Neon$4.41
- GPT-6.2 Astra$6.32
- GPT-6.2 Sol$16.87
- GPT-6 Astra$17.20
- Claude Fable 5.2$115.69
- Claude Sonnet 5.5$263.73
- Claude Opus 5.5$1376.70