Best LLM for agents: AA Agentic Index ranking
- models measured
- 87
- lead over #2
- +1.4
- median
- 18.5
Leader on AA Agentic Index · read October 20
As of October 20, 2026, Anthropic Claude Fable 5.1 leads agents with 57.9 on AA Agentic Index, ahead of Anthropic Claude Opus 5 (56.5) and Qwen3.8 Max (0902) (56.0). This ranking is one public board, shown as published, and is not part of the Vibe Coding Index.
| Rank | Model | AA Agentic Index | Evidence | Coding index | Output $/Mtok |
|---|---|---|---|---|---|
| 1 | Claude Fable 5.1 Anthropic | 57.9 | 89.1 | $50.00 | |
| 2 | Claude Opus 5 Anthropic | 56.5 | 81.8 | $25.00 | |
| 3 | Qwen3.8 Max (0902) Qwen | 56.0 | 63.5 | $6.00 | |
| 4 | Muse Spark 1.3 Meta | 55.5 | 79.7 | $4.25 | |
| 5 | GLM 5.3 Z.ai | 53.1 | 73.1 | $1.40 | |
| 6 | Grok 4.6 xAI | 53.0 | 76.4 | $6.00 | |
| 7 | GPT-6 Astra OpenAI | 51.0 | 88.7 | $50.00 | |
| 8 | GLM 5.3 Flash Z.ai | 50.9 | 71.0 | $0.16 | |
| 9 | Claude Fable 5 Anthropic | 50.7 | 82.1 | $50.00 | |
| 10 | GPT-5.6 Sol OpenAI | 50.2 | 76.7 | $20.00 | |
| 11 | Qwen3.8 2.4T A95B Qwen | 50.1 | Not benchmarked | $6.00 | |
| 12 | Kimi K3 Moonshot AI | 50.0 | 61.8 | $11.936 | |
| 13 | DeepSeek V4 Flash Vision Exp DeepSeek | 47.5 | 37.6 | $0.6468 | |
| 14 | Qwen3.8 27B Qwen | 45.8 | 73.6 | $2.20 | |
| 15 | Claude Sonnet 5 Anthropic | 43.6 | 63.5 | $10.00 | |
| 16 | Muse Spark 1.2 Meta | 43.2 | 65.1 | $4.25 | |
| 17 | GPT-5.6 Terra OpenAI | 43.2 | 69.0 | $12.00 | |
| 18 | GPT-5.6 Luna OpenAI | 42.1 | 61.5 | $1.20 | |
| 19 | Claude Opus 4.8 Anthropic | 41.9 | 66.0 | $25.00 | |
| 20 | DeepSeek V4 Pro 0813 DeepSeek | 41.3 | 54.6 | $1.98 | |
| 21 | Grok 4.5 xAI | 41.2 | 75.5 | $6.00 | |
| 22 | DeepSeek V4 Flash 0731 DeepSeek | 41.0 | Not benchmarked | $1.32 | |
| 23 | Gemini 3.8 Flash Google | 40.2 | 70.5 | $4.125 | |
| 24 | Claude Opus 4.7 Anthropic | 38.6 | 51.1 | $25.00 | |
| 25 | GLM 5.2 Z.ai | 38.4 | 57.7 | $2.42 | |
| 26 | GPT-5.5 OpenAI | 36.4 | 67.8 | $30.00 | |
| 27 | Claude Sonnet 4.6 Anthropic | 31.8 | 57.8 | $15.00 | |
| 28 | MiniMax M3 MiniMax | 29.5 | 57.8 | $1.20 | |
| 29 | Gemini 3.6 Flash Google | 29.0 | 48.3 | $3.75 | |
| 30 | Ling 3.0 Flash VL inclusionAI | 28.7 | Not benchmarked | $0.18 | |
| 31 | DeepSeek V4 Flash 0423 DeepSeek | 26.3 | 54.6 | $0.182 | |
| 32 | DeepSeek V4 Pro 0423 DeepSeek | 26.3 | 35.3 | $1.9105 | |
| 33 | Muse Spark 1.1 Meta | 25.8 | 48.9 | $4.25 | |
| 34 | Hy3 preview Tencent | 24.1 | 22.4 | $0.60 | |
| 35 | GLM 5.1 Z.ai | 23.9 | 44.0 | $3.036 | |
| 36 | T Inkling Small Thinking Machines | 23.5 | 29.9 | $1.20 | |
| 37 | Qwen3.7 Max Qwen | 22.5 | 45.1 | $4.425 | |
| 38 | T Inkling Thinking Machines | 22.5 | 49.7 | $4.05 | |
| 39 | MiMo-V2.5-Pro Xiaomi | 21.3 | 40.0 | $0.87 | |
| 40 | Kimi K2.7 Code Moonshot AI | 21.0 | 39.1 | $3.35 | |
| 41 | Kimi K2.6 Moonshot AI | 20.5 | 44.2 | $3.41 | |
| 42 | Nemotron 3 Ultra NVIDIA | 20.1 | Not benchmarked | $2.40 | |
| 43 | Ling 3.0 Flash inclusionAI | 19.3 | Not benchmarked | $0.063 | |
| 44 | Qwen3.6 27B Qwen | 18.5 | Not benchmarked | $2.00 | |
| 45 | GPT-5.4 Mini OpenAI | 17.9 | 28.2 | $4.50 | |
| 46 | Qwen3.7 Plus Qwen | 17.5 | Not benchmarked | $1.28 | |
| 47 | GPT-5.4 Nano OpenAI | 16.0 | Not benchmarked | $1.25 | |
| 48 | Claude Sonnet 4.5 Anthropic | 15.8 | 59.1 | $15.00 | |
| 49 | MiMo-V2.5 Xiaomi | 15.8 | 34.0 | $0.28 | |
| 50 | Grok 4.3 xAI | 15.5 | 22.5 | $2.50 | |
| 51 | MiniMax M2.7 MiniMax | 15.3 | 28.4 | $1.20 | |
| 52 | Gemini 3.5 Flash Lite Google | 14.3 | 33.6 | $2.50 | |
| 53 | LongCat 2.0 Meituan | 14.0 | Not benchmarked | $1.20 | |
| 54 | Qwen3.6 35B A3B Qwen | 13.1 | Not benchmarked | $0.95 | |
| 55 | Muse Glimmer 30B Meta | 8.5 | Not benchmarked | $1.20 | |
| 56 | Gemini 3.1 Pro Preview Google | 8.2 | 35.3 | $12.00 | |
| 57 | Claude Haiku 4.5 Anthropic | 8.0 | 49.8 | $5.00 | |
| 58 | Mistral Medium 3.5 Mistral AI | 7.0 | 9.2 | $7.50 | |
| 59 | DeepSeek V3.1 Terminus DeepSeek | 6.8 | Not benchmarked | $1.00 | |
| 60 | Gemma 4 31B Google | 4.2 | 38.3 | $0.40 | |
| 61 | gpt-oss-120b OpenAI | 3.7 | 34.5 | $0.60 | |
| 62 | Nemotron 3.5 Lightning NVIDIA | 3.5 | Not benchmarked | $0.20 | |
| 63 | Gemma 4 26B A4B Google | 2.9 | 22.8 | $0.255 | |
| 64 | Devstral 2 2512 Mistral AI | 2.3 | 53.8 | $2.00 | |
| 65 | Mistral Medium 3.1 Mistral AI | 1.9 | Not benchmarked | $2.00 | |
| 66 | Nemotron 3 Super NVIDIA | 1.7 | Not benchmarked | $0.45 | |
| 67 | Gemini 2.5 Pro Google | 1.6 | 37.0 | $10.00 | |
| 68 | 1.6 | 8.1 | $1.50 | ||
| 69 | Mercury 2 Inception | 1.6 | 0.0 | $0.75 | |
| 70 | Solar Pro 3 Upstage | 1.4 | Not benchmarked | $0.60 | |
| 71 | Granite 4.2 8B IBM Granite | 1.3 | Not benchmarked | $0.25 | |
| 72 | 1.3 | Not benchmarked | $3.50 | ||
| 73 | Ministral 3 14B 2512 Mistral AI | 1.1 | Not benchmarked | $0.20 | |
| 74 | Trinity Large Thinking Arcee AI | 1.0 | 5.4 | $0.80 | |
| 75 | Command A Cohere | 1.0 | Not benchmarked | $10.00 | |
| 76 | Mistral Large 3 2512 Mistral AI | 1.0 | Not benchmarked | $1.50 | |
| 77 | Nemotron 3 Nano 30B A3B NVIDIA | 1.0 | Not benchmarked | $0.24 | |
| 78 | o3 Mini High OpenAI | 0.9 | Not benchmarked | $4.40 | |
| 79 | Qwen3 Coder Next Qwen | 0.9 | Not benchmarked | $0.80 | |
| 80 | DeepSeek V3 0324 DeepSeek | 0.8 | 42.0 | $1.14 | |
| 81 | Ministral 3 3B 2512 Mistral AI | 0.8 | Not benchmarked | $0.10 | |
| 82 | Mistral Small 4 Mistral AI | 0.8 | Not benchmarked | $0.60 | |
| 83 | Llama 4 Maverick Meta Llama | 0.6 | 18.1 | $0.6525 | |
| 84 | Ministral 3 8B 2512 Mistral AI | 0.6 | Not benchmarked | $0.15 | |
| 85 | Llama 4 Scout Meta Llama | 0.5 | 9.1 | $0.30 | |
| 86 | Gemma 3 12B Google | 0.1 | Not benchmarked | $0.15 | |
| 87 | Gemma 3 27B Google | 0.1 | 4.9 | $0.35 |
What this measures
Can the model carry a task over several steps, calling tools along the way?
The agentic part of the Artificial Analysis evaluations: tasks where the model acts over several steps with tools, combined into one score from 0 to 100.
Read from Artificial Analysis (via OpenRouter) on October 20, 2026. Each model appears once with its best published setting; rows that match no model in our catalog are left out. AA Agentic Index benchmark page