Performance ranking
Top 10 included models · higher is better
Loading chartModel price and performance
GPQA Diamond21 matched evaluations · mean score · max reasoning setting
Loading chartPrice versus performance
Same test · fixed token workload
Loading chart1M input + 250k output tokens at current list prices. Log(1 + USD) cost axis; score axis zoomed. Dashed line: best included trade-offs.
What the workload costs
Input and output shown separately · 1M input + 250k output tokens
Loading chartSame input and output volumes for every model. No caching, batch discount or evaluation-token estimate is assumed. Listed prices may differ from the final routed bill.
Different routes to stronger models
How the release frontier moved
ECI index points · model release date · 227 attributed releases since 2023Loading chartCurrent ECI estimates by release date · not archived historical scores
What does a matching score cost today
GPQA Diamond · 1M input + 250k output tokens · 21 priced evaluationsLoading chartCurrent list prices · blank means unavailable · no caching or batch discounts
Prices we can trace back to the source
Selected dated offerings · USD per million tokens · uncached input
Loading chartSelected offerings, not a continuous price index or equal-quality comparison. Claude 3 Haiku’s price was announced before availability. DeepSeek V3 uses the announced post-promotion tariff from 8 February 2025.
Compare every benchmark
Read each test separately. Missing evaluations remain blank.
| Model | GPQA Diamond% mean score | FrontierMath tiers 1–3 · v2% mean score | MATH level 5% mean score | SimpleQA Verified% mean score | OTIS Mock AIME 2024–2025% mean score |
|---|---|---|---|---|---|
| DeepSeek V4.1 Flash 1017 | 93.2 | 71.2 | — | 42.5 | 99.4 |
| DeepSeek V4.1 Flash | 89.8 | 67.4 | — | 42.9 | 98.3 |
| Claude Haiku 5.5 | 89.6 | 75.1 | — | 23.8 | — |
| GPT-6.1 Sol | 95.4 | 93.7 | — | 73.9 | 100.0 |
| Claude Sonnet 5.5 | 95.6 | 88.8 | — | 46.5 | 100.0 |
| GPT-6 Luna | 90.5 | 78.9 | — | 41.4 | 98.9 |
| GPT-6 Sol | 94.3 | 89.8 | — | 60.7 | 100.0 |
| Claude Opus 5.5 | 90.6 | 91.2 | — | 72.2 | 100.0 |
| GPT-6 Astra | 95.8 | 93.7 | — | 75.6 | 100.0 |
| DeepSeek V4 Pro 0813 | 91.7 | 64.6 | — | 52.9 | 98.6 |
| Claude Opus 4.6 | 88.4 | 66.0 | — | 47.0 | 91.1 |
| Claude Sonnet 5 | 80.3 | 65.6 | — | 33.7 | 80.0 |
| Claude Sonnet 4.6 | 78.8 | — | — | 32.8 | 71.1 |
| Claude Opus 4.7 | 86.4 | 70.2 | — | — | 86.7 |
| Claude Fable 5 | 85.9 | 87.0 | — | — | 99.7 |
| DeepSeek V4 Flash 0731 | 91.0 | 57.5 | — | 33.6 | 94.4 |
| Claude Opus 5 | 93.9 | 85.6 | — | 59.9 | 98.9 |
| GPT-5.6 Terra | 93.3 | 86.0 | — | 43.2 | 99.7 |
| GPT-5.6 Luna | 91.6 | 82.1 | — | 41.0 | 98.3 |
| GPT-5.6 Sol | 93.5 | 89.1 | — | 69.7 | 100.0 |
| Claude Opus 4.8 | 91.0 | 80.0 | — | 53.0 | 98.3 |
| Muse Spark 1.3 | — | 74.0 | — | — | — |
| Claude Fable 5.1 | — | 90.2 | — | 70.8 | 100.0 |
