Start with a shortlist
Leaderboard
83 ranked models| Rank | Model | Compare | ||||||
|---|---|---|---|---|---|---|---|---|
| 01 |  Claude Opus 5.5Max, Default Fallback | 57.6 | 66.9% | 59.6% | $5.98 | 97 | 1M | |
| 02 | Claude Sonnet 5.5Max, Default Fallback | 56.0 | 61.0% | 63.6% | $5.46 | 141 | 1M | |
| 03 |  Muse Wonder FullMax | 55.8 | 62.6% | 57.1% | —Not measured | —Not measured | 1M | |
| 04 | Claude Fable 5.1Max, Default Fallback | 53.4 | 63.1% | 52.0% | $7.63 | 71 | 1M | |
| 05 |  GPT-6 AstraMax | 52.7 | 56.5% | 59.1% | $3.26 | 57 | 1.05M | |
| 06 |  Gemini 4 ArgonHigh | 52.6 | 61.8% | 57.1% | $1.99 | 100 | 1.049M | |
| 07 | Muse WonderMax | 52.4 | 60.3% | 52.0% | $4.52 | 181 | 1.049M | |
| 08 | GPT-6.1 SolMax | 51.8 | 54.2% | 56.1% | $0.72 | 66 | 1.05M | |
| 09 | Muse Spark 1.3Max | 48.1 | 58.8% | 33.3% | $1.60 | 134 | 1M | |
| 10 |  Grok 4.7Xhigh | 46.4 | 57.4% | 25.8% | $3.74 | 93 | 500K |
Artificial Analysis measurements, each model at its highest-scoring measured setting. Cost is the Index evaluation per task; speed is median output for a 10K-token prompt. Methodology