Which model should I use?

Pick the job, set what you're willing to pay, and see what the measured data says. Capability comes from human preference votes on that specific kind of work; price is live API pricing. Nothing here is a blended "intelligence score" we invented.

Required by every agent framework

44 models measured · Arena text / coding

Rows on the frontier are Pareto-optimal: nothing we track is both stronger and cheaper. "Strength vs leader" rescales the board so 100% = parity with the best model on it.

#ModelProviderScoreStrength$/1MContextToolsFrontier
1 Gemini 4 Carbon
google/gemini-4-carbon
Google 1564 100% $4.001.048576M tools frontier
2 Gemini 4 Neon
google/gemini-4-neon
Google 1555 97% $6.001.048576M tools
3 Claude Opus 5.5
anthropic/claude-opus-5.5
Anthropic 1543 94% $8.001M tools
4 Gemini 4 Argon
google/gemini-4-argon
Google 1539 93% $4.001.048576M tools frontier
5 Claude Opus 5
anthropic/claude-opus-5
Anthropic 1532 91% $10.001M tools
6 Claude Fable 5.1
anthropic/claude-fable-5.1
Anthropic 1527 89% $20.001M tools
7 Claude Fable 5
anthropic/claude-fable-5
Anthropic 1519 87% $20.001M tools
8 Gemini 4 Hydrogen
google/gemini-4-hydrogen
Google 1512 85% $3.001.048576M tools frontier
9 Gemini 3.8 Flash
google/gemini-3.8-flash
Google 1509 84% $1.661.048576M tools frontier
10 GLM 5.3 Flash
z-ai/glm-5.3-flash
Z.ai (Zhipu) 1509 84% $0.051.048576M tools frontier
11 Kimi K3
moonshotai/kimi-k3
Moonshot AI 1505 83% $4.141.048576M tools
12 Qwen3.8 Max (0902)
qwen/qwen3.8-max-0902
Qwen (Alibaba) 1503 83% $3.001M tools
13 DeepSeek V4.1 Flash
deepseek/deepseek-v4.1-flash
DeepSeek 1503 83% $0.091.048576M tools
14 GLM 5.3
z-ai/glm-5.3
Z.ai (Zhipu) 1501 82% $0.401.048576M tools
15 Claude Sonnet 5.5
anthropic/claude-sonnet-5.5
Anthropic 1500 82% $4.001M tools
16 GPT-6.1 Sol
openai/gpt-6.1-sol
OpenAI 1490 79% $4.001.05M tools
17 Kimi K2.6
moonshotai/kimi-k2.6
Moonshot AI 1489 79% $1.34262K tools
18 GLM 5.2
z-ai/glm-5.2
Z.ai (Zhipu) 1488 78% $1.071.048576M tools
19 Claude Fable 5.2
anthropic/claude-fable-5.2
Anthropic 1487 78% $16.001M tools
20 GLM 5.1
z-ai/glm-5.1
Z.ai (Zhipu) 1485 78% $1.83205K tools
21 Kimi K2.5
moonshotai/kimi-k2.5
Moonshot AI 1484 77% $1.00262K tools
22 Kimi K3.1 Pro
moonshotai/kimi-k3.1-pro
Moonshot AI 1484 77% $2.391.048576M tools
23 Claude Sonnet 5
anthropic/claude-sonnet-5
Anthropic 1483 77% $4.001M tools
24 GPT-6.1 Terra
openai/gpt-6.1-terra
OpenAI 1482 77% $2.001.05M tools
25 Qwen3.8 27B
qwen/qwen3.8-27b
Qwen (Alibaba) 1481 77% $0.411.024M tools
26 Grok 4.5
x-ai/grok-4.5
xAI 1474 75% $3.00500K tools
27 DeepSeek V4 Pro 0813
deepseek/deepseek-v4-pro-0813
DeepSeek 1469 73% $1.981.048576M tools
28 DeepSeek V4 Pro 0423
deepseek/deepseek-v4-pro
DeepSeek 1469 73% $1.051.048576M tools
29 MiniMax M3
minimax/minimax-m3
MiniMax 1467 73% $0.531.048576M tools
30 Grok 4.6
x-ai/grok-4.6
xAI 1464 72% $3.00500K tools
31 Mistral Medium 3.5
mistralai/mistral-medium-3-5
Mistral 1461 71% $3.00262K tools
32 Grok 4.20
x-ai/grok-4.20
xAI 1459 71% $1.562M tools
33 Grok 4.20 Multi-Agent
x-ai/grok-4.20-multi-agent
xAI 1458 70% $1.562M —
34 DeepSeek V4 Flash 0731
deepseek/deepseek-v4-flash-0731
DeepSeek 1457 70% $0.101.048576M tools
35 DeepSeek V4 Flash 0423
deepseek/deepseek-v4-flash
DeepSeek 1457 70% $0.111.048576M tools
36 DeepSeek V3.2
deepseek/deepseek-v3.2
DeepSeek 1454 69% $0.29164K tools
37 MiniMax M2.7
minimax/minimax-m2.7
MiniMax 1454 69% $0.90197K tools
38 Claude Haiku 5.5
anthropic/claude-haiku-5.5
Anthropic 1452 69% $0.201M tools
39 GPT-6.2 Astra
openai/gpt-6.2-astra
OpenAI 1450 68% $16.001.05M tools
40 GPT-6.2 Sol
openai/gpt-6.2-sol
OpenAI 1440 66% $4.001.05M tools
41 Grok 4.7
x-ai/grok-4.7
xAI 1423 62% $3.00500K tools
42 Grok 4.3
x-ai/grok-4.3
xAI 1414 59% $1.561M tools
43 MiniMax M2.5
minimax/minimax-m2.5
MiniMax 1381 52% $0.44198K tools
44 Ministral 3 8B 2512
mistralai/ministral-8b-2512
Mistral 1202 22% $0.15262K tools

Capability: Arena leaderboard dataset (CC-BY-4.0), published 2026-11-19. Pricing: OpenRouter, retrieved 2026-11-22. Blended price = (input×3 + output×1) ÷ 4.

For agents: this whole decision table is one fetch — GET /api/v1/route/{task}.json, or /api/v1/route/all.json for every task at once. No key, open CORS. See API & Bots.

How to read this honestly

  • Capability here is human preference, not correctness. Arena scores say which answer people preferred in head-to-head votes on that category. A model can be preferred and still be wrong.
  • Elo boards and agent boards use different scales. On the Elo boards a dead tie with the leader is a 0.50 win probability; on the Bradley-Terry agent boards parity is 1.0. We rescale both to "% of leader strength" for display and publish the raw numbers in the API.
  • Models with no score on a board are left out of that task rather than given an estimated one.
  • Price is OpenRouter's listed price at retrieval time; going direct to a provider can be cheaper or dearer.