Claude Code最佳AI模型
Claude Code is Anthropic's CLI agent for software engineering. It excels at multi-file editing, test writing, and complex refactoring. Benefits from large context windows and strong reasoning.
Claude Code的重要因素
Best Models for Claude Code
Top 15 by tool-optimized score
Claude Code全部模型排名（440个模型）
Scored by: benchmark performance (90%) from MMLU, GPQA, HumanEval, SWE-bench, and 15+ standardized evaluations, with capabilities and context as tiebreakers (10%).
| # | 模型 | 评分 | 输出$/百万 |
|---|---|---|---|
| 1 | Muse Spark 1.1 Arena Elo: 1491 | 87 | $4.25 |
| 2 | MiMo-V2.6-Pro Arena Elo: 1480 | 86 | $0.870 |
| 3 | Gemini 3.5 Flash Arena Elo: 1477 | 86 | $9.00 |
| 4 | Claude Opus 5.5 | 85 | $20.00 |
| 5 | Claude Fable 5.1 | 85 | $50.00 |
| 6 | Claude Fable 5.1 (batch) | 85 | $25.00 |
| 7 | GLM 5.3 Flash Arena Elo: 1474 | 85 | $0.500 |
| 8 | Claude Opus 5 | 85 | $25.00 |
| 9 | Grok 4.5 Arena Elo: 1465 | 85 | $6.00 |
| 10 | Claude Fable 5 SWE-bench: 95% | 85 | $50.00 |
| 11 | Claude Fable 5 (batch) | 85 | $25.00 |
| 12 | Claude Opus 4.8 (batch) | 85 | $12.50 |
| 13 | MiMo-V2.5-Pro Arena Elo: 1467 | 85 | $0.870 |
| 14 | Claude Opus 4.7 (batch) | 85 | $12.50 |
| 15 | GLM 5.1 Arena Elo: 1465 | 85 | $3.03 |
| 16 | MiMo-V2.6-Flash Arena Elo: 1454 | 84 | $0.280 |
| 17 | Gemini 3.5 Flash Lite Arena Elo: 1456 | 84 | $2.50 |
| 18 | Hy3 Arena Elo: 1457 | 84 | $0.528 |
| 19 | Qwen3.7 Plus Arena Elo: 1456 | 84 | $1.28 |
| 20 | Qwen3.6 Max Preview Arena Elo: 1460 | 84 | $6.16 |
| 21 | GPT-5.5 Pro (batch) | 84 | $90.00 |
| 22 | GPT-5.5 (batch) | 84 | $15.00 |
| 23 | Kimi K2.6 Arena Elo: 1461 | 84 | $4.00 |
| 24 | Gemma 4 31B Arena Elo: 1453 | 84 | $0.340 |
| 25 | Gemini 3.1 Pro Preview Custom Tools | 84 | $12.00 |
| 26 | Gemini 3.1 Pro Preview (batch) | 84 | $6.00 |
| 27 | GLM 5 Arena Elo: 1457 | 84 | $1.92 |
| 28 | Inkling Arena Elo: 1442 | 83 | $4.05 |
| 29 | MiniMax M3 Arena Elo: 1440 | 83 | $1.20 |
| 30 | Qwen3.6 Plus Arena Elo: 1443 | 83 | $1.95 |
| 31 | GPT-5.4 Pro | 83 | $180.00 |
| 32 | GPT-5.4 Pro (batch) | 83 | $90.00 |
| 33 | GPT-5.4 (batch) | 83 | $7.50 |
| 34 | GPT-5.3-Codex | 83 | $14.00 |
| 35 | Qwen3.5 397B A17B Arena Elo: 1442 | 83 | $3.50 |
| 36 | Claude Opus 4.6 (batch) | 83 | $12.50 |
| 37 | GPT-5.2-Codex | 83 | $14.00 |
| 38 | GLM 4.7 Arena Elo: 1441 | 83 | $2.20 |
| 39 | GPT-5.2 Pro | 83 | $168.00 |
| 40 | GPT-5.2 Pro (batch) | 83 | $84.00 |
| 41 | GPT-5.2 (batch) | 83 | $7.00 |
| 42 | Claude Opus 4.1 Arena Elo: 1450 | 83 | $75.00 |
| 43 | Grok 4.7 | 82 | $4.80 |
| 44 | Qwen3.8 27B Arena Elo: 1438 | 82 | $3.00 |
| 45 | Grok 4.6 | 82 | $6.00 |
| 46 | GPT-5.6 Luna Pro | 82 | $1.20 |
| 47 | GPT-5.6 Luna Pro (batch) | 82 | $0.600 |
| 48 | GPT-5.6 Luna | 82 | $1.20 |
| 49 | GPT-5.6 Luna (batch) | 82 | $0.600 |
| 50 | GPT-5.6 Terra Pro | 82 | $12.00 |
Based on our analysis of coding benchmarks, capability matching, and pricing, Muse Spark 1.1 currently ranks #1 for Claude Code. Rankings are rebuilt as benchmark, pricing, and provider data refresh.
We score models using benchmark performance (90%) from LMArena, HumanEval, SWE-bench, MMLU, and 15+ standardized evaluations. Capabilities and context serve as tiebreakers (10%). Only models with the capabilities Claude Code needs are included in the tool-specific rankings.
We currently track 440 AI models compatible with Claude Code. This includes models from OpenAI, Anthropic, Google, DeepSeek, and other providers accessible via API.
Many open-source models are compatible with Claude Code through API providers like OpenRouter, Together AI, and Groq. Check our rankings to see which open-source models perform best.
Rankings refresh whenever the underlying benchmark, pricing, and catalog sources refresh. That means some signals update faster than others, and the page reflects the latest verified source data available.