| # | model | served by | intelligence index | share of leader | in / out per M |
|---|---|---|---|---|---|
| 01 |  claude-opus-5-5 | Anthropic | 58.4 | $4.00 / $20.00 | |
| 02 |  gpt-6-astra | OpenAI | 56.9 | $10.00 / $50.00 | |
| 03 | gpt-6.1-sol | OpenAI | 54.5 | $2.00 / $10.00 | |
| 04 | claude-sonnet-5-5 | Anthropic | 54.2 | $2.00 / $10.00 | |
| 05 | gpt-6.1-terra | OpenAI | 51.0 | $1.00 / $5.00 | |
| 06 |  gemini-4-argon | Google Vertex AI | 50.4 | $2.00 / $10.00 | |
| 07 | claude-fable-5.1 | Anthropic | 49.7 | $10.00 / $50.00 | |
| 08 |  kimi-k3.1-pro | Moonshot AI | 46.0 | $2.00 / $10.00 | |
| 09 |  muse-spark-1.3 | meta | 36.4 | $1.25 / $4.25 | |
| 10 | claude-haiku-5-5 | Anthropic | 36.2 | $0.10 / $0.50 | |
| 11 |  grok-4.7 | xAI | 35.0 | $2.00 / $6.00 | |
| 12 |  qwen3.8-max | Alibaba Cloud | 33.6 | $2.00 / $6.00 | |
| 13 |  deepseek-v4.1-flash-1017 | DeepSeek | 32.5 | $0.30 / $1.20 | |
| 14 |  glm-5.3 | Z.ai | 31.7 | $1.40 / $4.40 | |
| 15 | gemini-3.8-flash | Google Vertex AI | 30.5 | $0.75 / $3.75 | |
| 16 |  mimo-v2.6-pro | DeepInfra | 29.9 | $0.43 / $0.87 | |
| 17 |  minimax-m3.1-flash | MiniMax | 28.9 | $0.15 / $0.60 | |
| 18 | step-5-preview | DeepInfra | 28.5 | $1.00 / $2.70 | |
| 19 | gpt-6-luna | OpenAI | 27.8 | $0.10 / $0.50 | |
| 20 | glm-5.3-flash | Z.ai | 27.8 | $0.15 / $0.50 | |
| 21 | mimo-v2.6-flash | DeepInfra | 26.7 | $0.14 / $0.28 | |
| 22 |  mistral-large-4 | Mistral AI | 26.0 | $0.68 / $2.09 | |
| 23 |  muse-spark-open | Together AI | 24.7 | $1.25 / $4.25 | |
| 24 | deepseek-v4-pro-0813 | DeepSeek | 23.1 | $1.32 / $3.96 | |
| 25 | qwen3.8-27b | DeepInfra | 20.1 | $0.20 / $2.50 | |
| 26 | minimax-m3 | MiniMax | 18.7 | $0.30 / $1.20 | |
| 27 | kimi-k3 | Moonshot AI | 17.4 | $3.00 / $15.00 | |
| 28 |  inkling | Thinking Machines | 12.5 | $1.87 / $4.68 | |
| 29 |  nvidia-nemotron-3-ultra | Doubleword | 11.8 | $0.50 / $2.50 | |
| 30 |  gemini-3.5-flash-lite | Google Gemini API | 11.7 | $0.30 / $2.50 |
Bars are each score's share of the leader's, so a short bar is a real gap and not a rounding difference. One row per model family, using the lab's own endpoint where it exists.
claude-opus-5-5 leads gpt-6-astra by 1.5 points
One bar per ranked model, best on the left, drawn as a share of the leader. 4 models sit within 10 percent of the leader, so the top of this list is a cluster rather than a winner. The dashed line is the median.
Smartest overall 30 ranked by intelligence index
The Artificial Analysis Intelligence Index combines many independent evaluations (reasoning, coding, math, science, agentic tool use) into a single capability score. It is the best at-a-glance signal for raw model intelligence, but always validate on your own workload, since the best model for a specific task is often not the highest overall.
method
Scores for Intelligence Index come from Artificial Analysis, an independent benchmarking service. When a model is served by several providers (Anthropic direct, AWS Bedrock, Google Vertex), one canonical entry represents the model family so the ranking is not padded with duplicates. Benchmarks measure specific skills: validate on your own workload before committing.
Built from the same catalog the router reads at request time, rebuilt daily.
other lists 8
one api for every model on this list
Requesty is OpenAI-compatible. Switch between any two models above by changing one parameter, and keep the failover, caching and spend controls.