Best LLM for following instructions
- models ranked
- 156
- lead over #2
- +3 Elo
LMArena Text: instruction following · read October 22
As of October 22, 2026, Anthropic Claude Opus 5.5 leads 156 models on LMArena Text: instruction following with 1537, ahead of Google Gemini 4 Argon (1534) and Anthropic Claude Opus 4.6 (1521).
The full best LLM for general use ranking combines this benchmark with others.
| Rank | Model | LMArena Text: instruction following | Evidence | Coding index | Output $/Mtok |
|---|---|---|---|---|---|
| 1 | Claude Opus 5.5 Anthropic | 1537 | 88.8 | $20.00 | |
| 2 | Gemini 4 Argon Google | 1534 | 81.0 | $10.00 | |
| 3 | Claude Opus 4.6 Anthropic | 1521 | 57.2 | $25.00 | |
| 4 | Claude Fable 5.1 Anthropic | 1518 | 84.2 | $50.00 | |
| 5 | Claude Opus 5 Anthropic | 1518 | 83.0 | $25.00 | |
| 6 | Claude Fable 5 Anthropic | 1502 | 79.3 | $50.00 | |
| 7 | Claude Opus 4.7 Anthropic | 1498 | 67.5 | $25.00 | |
| 8 | Claude Sonnet 5.5 Anthropic | 1492 | 87.3 | $10.00 | |
| 9 | Gemini 3.8 Flash Google | 1492 | 61.6 | $3.75 | |
| 10 | MiMo-V2.6-Pro Xiaomi | 1491 | 65.4 | $0.87 | |
| 11 | Muse Wonder Meta | 1491 | 71.9 | $20.00 | |
| 12 | Kimi K3 Moonshot AI | 1485 | 62.3 | $15.00 | |
| 13 | Gemini 3.7 Flash Google | 1484 | 58.0 | $3.75 | |
| 14 | GPT-5.6 Sol OpenAI | 1483 | 73.5 | $20.00 | |
| 15 | Qwen3.8 Max (0902) Qwen | 1481 | 62.4 | $6.00 | |
| 16 | GLM 5.3 Z.ai | 1479 | 70.6 | $4.40 | |
| 17 | Muse Spark 1.3 Meta | 1479 | 57.4 | $4.25 | |
| 18 | GPT-5.5 OpenAI | 1478 | 71.3 | $30.00 | |
| 19 | Claude Opus 4.5 Anthropic | 1478 | 44.0 | $25.00 | |
| 20 | GLM 5.3 Flash Z.ai | 1478 | 57.7 | $0.50 |
How this list is built
Generated from the dataset on every refresh. Nobody picks the order by hand.
Models by their LMArena Text: instruction following result, as the board publishes it, best first. Prompts with explicit instructions the answer has to follow. People ask a question, compare two anonymous answers and vote for the better one; the rating is built from those votes, as LMArena publishes it.
- OpenRouter, fetched
- LMArena (Hugging Face dataset), fetched