Artificial Analysis Coding Index (Maximum Reasoning)
22 models
Top 5 Models Performance
| Anthropic: Claude Fable 5 | ######################################## | 62 |
| Xiaomi: MiMo-V2.5 Pro | ####################################### | 60.2 |
| Anthropic: Claude Opus 4.8 | ##################################### | 56.7 |
| OpenAI: gpt-5.3-codex | ################################## | 53.1 |
| OpenAI: gpt-5.2 | ############################### | 48.7 |
| Rank | Model | Score | |
|---|---|---|---|
| 🇺🇸 | 1 | Anthropic: Claude Fable 5 | 62 |
| 🇨🇳 | 2 | Xiaomi: MiMo-V2.5 Pro小米集团 | 60.2 |
| 🇺🇸 | 3 | Anthropic: Claude Opus 4.8 | 56.7 |
| 🇺🇸 | 4 | OpenAI: gpt-5.3-codex | 53.1 |
| 🇺🇸 | 5 | OpenAI: gpt-5.2 | 48.7 |
| 🇺🇸 | 6 | OpenAI: gpt-5.2-codex | 43 |
| 🇺🇸 | 7 | SpaceXAI: Grok 4.3 | 42.2 |
| 🇫🇷 | 8 | Mistral AI: mistral-medium-3.5-128b | 35.4 |
| 🇺🇸 | 9 | SpaceXAI: Grok 4.1 Fast | 30.9 |
| 🇺🇸 | 10 | SpaceXAI: Grok 4 Fast | 27.4 |
| 🇺🇸 | 11 | SpaceXAI: grok-code-fast-1 | 23.7 |
| 🇺🇸 | 12 | Google: Gemma 4 26B A4B Instruct | 22.4 |
| 🇺🇸 | 13 | OpenAI: o1 | 20.5 |
| ❓ | 14 | NousResearch: hermes-4-405b | 16 |
| 🇨🇳 | 15 | DeepSeek: DeepSeek R1深度求索 | 15.9 |
| 🇺🇸 | 16 | Meta Llama: Llama 4 Maverick Instruct | 15.6 |
| ❓ | 17 | NousResearch: hermes-4-70b | 14.4 |
| 🇺🇸 | 18 | OpenAI: GPT-3.5 Turbo | 10.7 |
| 🇺🇸 | 19 | Meta Llama: Llama 3.3 70B Instruct | 10.7 |
| 🇦🇪 | 20 | Technology Innovation Institute: falcon-h1r-7b | 9.8 |
| 🇺🇸 | 21 | Google: Gemma 4 E2B Instruct | 9 |
| 🇨🇳 | 22 | Qwen: Qwen3 0.6B通义千问 | 0.9 |