Best LLM
Which LLM writes the best text?
In LMArena Text, Gemini 4 Argon is ahead with a rating of 1,522.
The ranking rests on this one measurement alone, a rating from publicly collected pairwise votes, and is not an area of the Gradually overall index.
Among the ten best, Gemini 3.8 Flash is the cheapest at $1.50 per million tokens.
The fastest among them is Muse Spark 1.3, at 102 tokens per second.
- Measured models
- 80
- Data as of
- Nov 10, 2026
Ranking: Writing
The ranking rests on LMArena Text alone, a rating from publicly collected pairwise votes run by LMArena. The data contains no dedicated test for writing, more in the methodology.
| Rank | Model | LMArena Text | Overall index | Price (3:1) | Speed |
|---|---|---|---|---|---|
| 1 | Gemini 4 Argon | 1,522 | 121.4 | $4 | 46 |
| 2 | Muse Wonder | 1,511 | 113.4 | $8 | n/a |
| 3 | Claude Opus 5.5 | 1,506 | 125.7 | $8 | 75 |
| 4 | Claude Fable 5 | 1,504 | 112.8 | $20 | 50 |
| 5 | Claude Fable 5.1 | 1,503 | 124.1 | $20 | 59 |
| 6 | Gemini 3.8 Flash | 1,497 | 106.8 | $1.50 | 81 |
| 7 | Muse Spark 1.3 | 1,496 | 107.3 | $2 | 102 |
| 8 | Gemini 3.7 Flash | 1,489 | 107.9 | n/a | 72 |
| 8 | Claude Opus 5 | 1,489 | 117.4 | $10 | 71 |
| 10 | Kimi K3 | 1,487 | 101.1 | $6 | 61 |
| 10 | Gemini 3.1 Pro Preview | 1,487 | 91.4 | $4.50 | 96 |
| 12 | GPT-5.6 Sol | 1,485 | 109.1 | $8 | 75 |
| 13 | Gemini 3.6 Flash | 1,483 | 97.6 | $1.50 | 110 |
| 13 | GLM-5.3 | 1,483 | 100.6 | $2.15 | 81 |
| 15 | GPT-6.1 Sol | 1,482 | 118.0 | $4 | 44 |
| 16 | Gemini 3.5 Flash | 1,478 | 97.8 | n/a | 142 |
| 17 | GLM-5.2 | 1,475 | 89.0 | $2.15 | 73 |
| 17 | DeepSeek-V4.1-Flash | 1,475 | 99.5 | $0.26 | 118 |
| 19 | GLM-5.3-Flash | 1,474 | 94.3 | $0.24 | 101 |
| 19 | GPT-6 Astra | 1,474 | 120.2 | $20 | 58 |
Show all 80 models
| Rank | Model | LMArena Text | Overall index | Price (3:1) | Speed |
|---|---|---|---|---|---|
| 21 | Claude Sonnet 5.5 | 1,472 | 121.8 | $4 | 103 |
| 21 | Claude Sonnet 4.6 | 1,472 | 86.7 | $6 | 38 |
| 23 | GPT-5.6 Terra | 1,468 | 100.8 | $4.50 | 72 |
| 24 | MiMo-V2.5-Pro | 1,467 | 73.0 | $0.54 | 59 |
| 25 | Grok 4.5 | 1,466 | 94.8 | $3 | 52 |
| 26 | GLM-5.1 | 1,465 | 82.9 | $2.15 | 39 |
| 27 | Claude Sonnet 5 | 1,461 | 92.1 | $4 | 77 |
| 27 | Kimi K2.6 | 1,461 | 84.9 | $1.71 | 32 |
| 29 | GLM-5 | 1,457 | 72.7 | $1.55 | 41 |
| 29 | GPT-6 Sol | 1,457 | 109.1 | $4 | 82 |
| 31 | Grok 4.6 | 1,456 | 102.2 | $3 | 58 |
| 32 | Gemini 3.5 Flash-Lite | 1,455 | 73.7 | $0.85 | 145 |
| 33 | Hy3 | 1,454 | n/a | $0.25 | 36 |
| 34 | GPT-5.6 Luna | 1,453 | 93.8 | $0.45 | 123 |
| 35 | Gemma 4 31B | 1,452 | n/a | $0.21 | 43 |
| 36 | GPT-5.4 mini | 1,447 | 75.2 | $1.69 | 134 |
| 37 | Gemini 2.5 Pro | 1,446 | 61.4 | $3.44 | 74 |
| 38 | Inkling | 1,443 | 72.0 | $1.76 | n/a |
| 39 | Grok 4.7 | 1,442 | 103.3 | $3 | 47 |
| 39 | Grok 4.3 | 1,442 | 71.6 | $1.56 | 66 |
| 41 | GLM-4.7 | 1,441 | 56.3 | $0.78 | 25 |
| 41 | GPT-6 Luna | 1,441 | 95.8 | $0.20 | 77 |
| 43 | Gemma 4 26B A4B | 1,437 | n/a | $0.14 | 71 |
| 44 | MiniMax M3 | 1,436 | 83.0 | $0.52 | 82 |
| 45 | GPT-5 | 1,435 | 77.6 | $3.44 | 83 |
| 46 | MiMo-V2.5 | 1,434 | 69.2 | $0.18 | 55 |
| 47 | Mistral Medium 3.5 | 1,427 | 46.3 | $3 | 118 |
| 48 | DeepSeek-V3.2 | 1,425 | 48.6 | $0.29 | 31 |
| 49 | GLM-4.6 | 1,424 | 46.0 | $0.88 | 28 |
| 50 | Claude Haiku 5.5 | 1,422 | 102.8 | $0.20 | 164 |
| 50 | DeepSeek-V3.2 Exp | 1,422 | n/a | n/a | n/a |
| 52 | DeepSeek-V3.1 | 1,417 | n/a | $0.45 | 32 |
| 53 | MiniMax M2.7 | 1,415 | 65.1 | $0.37 | 45 |
| 53 | DeepSeek-V3.1 Terminus | 1,415 | n/a | $0.45 | 21 |
| 53 | Mistral Large 3 | 1,415 | 32.8 | $0.75 | 86 |
| 56 | GLM-4.5 | 1,411 | 48.9 | $1 | 28 |
| 57 | Gemini 2.5 Flash | 1,409 | n/a | $0.85 | n/a |
| 58 | Inkling Small | 1,405 | 68.7 | $0.52 | 61 |
| 59 | GPT-5.4 nano | 1,401 | 64.0 | $0.46 | 98 |
| 60 | DeepSeek-R1 | 1,398 | 44.6 | $1.15 | n/a |
| 61 | Step 3.5 Flash | 1,393 | n/a | $0.15 | 82 |
| 62 | MiniMax M2.5 | 1,391 | 61.6 | $0.52 | 49 |
| 63 | GPT-5 mini | 1,390 | 65.8 | $0.69 | 105 |
| 64 | Solar Pro 4 | 1,387 | 65.5 | $0.52 | 39 |
| 65 | GLM-4.6V | 1,378 | n/a | $0.45 | 27 |
| 66 | GLM-4.5 Air | 1,373 | n/a | $0.32 | 60 |
| 67 | Trinity Large Thinking | 1,367 | n/a | $0.39 | n/a |
| 68 | GLM-4.7-Flash | 1,364 | n/a | $0.15 | 44 |
| 68 | MiniMax M1 | 1,364 | n/a | $0.96 | 48 |
| 70 | DeepSeek-V3 | 1,358 | 19.8 | $0.37 | 40 |
| 71 | GLM-4.5V | 1,352 | n/a | $0.90 | 65 |
| 71 | GPT OSS 120B | 1,352 | 43.4 | $0.20 | 92 |
| 73 | MiniMax M2 | 1,344 | n/a | $0.45 | n/a |
| 74 | GPT-5 nano | 1,338 | 32.5 | $0.14 | 136 |
| 75 | Amazon Nova 2 Lite | 1,335 | n/a | $0.85 | 102 |
| 76 | GPT OSS 20B | 1,318 | 30.9 | $0.06 | 85 |
| 77 | Granite 4.1 8B | 1,304 | n/a | $0.063 | n/a |
| 78 | Command R+ | 1,262 | n/a | n/a | n/a |
| 79 | Phi-4 | 1,256 | n/a | $0.07 | 93 |
| 80 | Command R | 1,227 | n/a | n/a | n/a |
The 80 are the models with a value in LMArena Text. The overall index counts different sets of models, so the numbers differ between pages.
Price in dollars per million tokens, mixed from 3 parts input and 1 part output. Speed in tokens per second.
Which tests count here?
LMArena Text, Style Control
80 models tested · LMArena