Blind comparison of replies in French. It needs a collection of its own because the order of models changes noticeably from language to language — and the overall ranking does not show that difference.
| # | Model | Developer | Arena Score, French | Sign in$ / 1M | Output$ / 1M | Contexttokens |
|---|---|---|---|---|---|---|
| 1 | Claude Opus 5 | Anthropic | 1,522.24 1,479–1,565 | $5 | $25 | 1,000K |
| 2 | Claude Fable 5 ≈ | Anthropic | 1,519.66 1,494–1,545 | $10 | $50 | 1,000K |
| 3 | GPT-5.6 Terra ≈ | OpenAI | 1,503.02 1,465–1,542 | $2 | $12 | 1,050K |
| 4 | Muse Spark 1.1 ≈ | Meta AI | 1,501.23 1,470–1,533 | $1.25 | $4.25 | 1,049K |
| 5 | Claude Opus 4.7 ≈ | Anthropic | 1,499.99 1,484–1,516 | $5 | $25 | 1,000K |
| 6 | Claude Opus 4.6 ≈ | Anthropic | 1,497.74 1,483–1,513 | $5 | $25 | 1,000K |
| 7 | Gemini 3 Pro ≈ | Google DeepMind | 1,495.19 1,474–1,516 | $1.6 | $9.6 | 1,049K |
| 8 | Gemini 3 Flash ≈ | Google DeepMind | 1,491.08 1,468–1,514 | $0.4 | $2.4 | 1,049K |
| 9 | Gemini 3.6 Flash ≈ | Google DeepMind | 1,489.78 1,454–1,526 | $1.5 | $7.5 | 1,049K |
| 10 | MiMo V2.5 Pro ≈ | Xiaomi | 1,489.7 1,473–1,506 | $0.44 | $0.87 | 1,050K |
| 11 | Gemini 3.1 Pro Preview ≈ | Google DeepMind | 1,488.01 1,474–1,502 | $2 | $12 | 1,049K |
| 12 | ERNIE 5.1 ≈ | Baidu (Ernie) | 1,483.39 1,465–1,501 | $0.75 | $3 | 119K |
| 13 | Grok 4.20 (Reasoning) ≈ | xAI | 1,477.38 1,462–1,493 | $2 | $6 | 2,000K |
| 14 | Claude 4.5 Opus ≈ | Anthropic | 1,477.28 1,461–1,493 | $5 | $25 | 200K |
| 15 | GLM 5.2 ≈ | Zhipu AI / Z.ai | 1,476.9 1,454–1,500 | $0.42 | $1.32 | 1,049K |
| 16 | Qwen3.7 Plus ≈ | Alibaba (Qwen) | 1,475.27 1,455–1,496 | $0.5 | $3 | 1,000K |
| 17 | GLM 5.1 ≈ | Zhipu AI / Z.ai | 1,474.78 1,456–1,494 | $0.06 | $0.22 | 205K |
| 18 | Claude Opus 4.8 ≈ | Anthropic | 1,471.76 1,453–1,490 | $5 | $25 | 1,000K |
| 19 | Gemini 2.5 Pro ≈ | Google DeepMind | 1,470.66 1,456–1,485 | $1.25 | $10 | 1,049K |
| 20 | Kimi K2.6 ≈ | Moonshot AI (Kimi) | 1,469.9 1,452–1,487 | $0.22 | $1.14 | 262K |
| 21 | Grok 4.20 Multi Agent Beta 0309 ≈ | xAI | 1,468.57 1,453–1,484 | $2 | $6 | 2,000K |
| 22 | MiMo V2 Pro ≈ | Xiaomi | 1,467.9 1,446–1,489 | $0.44 | $0.87 | 1,049K |
| 23 | Gemma 4 26B-A4B ≈ | Google DeepMind | 1,466.68 1,426–1,507 | $0.05 | $0.29 | 262K |
| 24 | Seed 2.0 Pro ≈ | ByteDance (Doubao) | 1,466.36 1,451–1,481 | $0.5 | $3 | 131K |
| 25 | GPT-5.6 Sol ≈ | OpenAI | 1,465.3 1,430–1,501 | $5 | $30 | 1,050K |
| 26 | Mistral Large 3 ≈ | Mistral AI | 1,465.23 1,448–1,483 | $0.5 | $1.5 | 256K |
| 27 | DeepSeek V4 Pro ≈ | DeepSeek | 1,465.07 1,449–1,481 | $0.44 | $0.87 | 1,049K |
| 28 | Claude Sonnet 4.6 ≈ | Anthropic | 1,464.7 1,449–1,481 | $3 | $15 | 1,000K |
| 29 | Claude Sonnet 5 ≈ | Anthropic | 1,464.66 1,438–1,491 | $2 | $10 | 1,000K |
| 30 | GLM 4.6 ≈ | Zhipu AI / Z.ai | 1,461.66 1,437–1,486 | $0.29 | $1.14 | 205K |
| 31 | Mistral Medium 3.1 ≈ | Mistral AI | 1,461.53 1,447–1,477 | $0.4 | $2 | 262K |
| 32 | Gemini 3.5 Flash ≈ | Google DeepMind | 1,460.6 1,427–1,494 | $1.5 | $9 | 1,049K |
| 33 | Qwen3 VL 235B A22B Instruct ≈ | Alibaba (Qwen) | 1,459.98 1,418–1,502 | $0.4 | $1.6 | 262K |
| 34 | Qwen3 Max Preview ≈ | Alibaba (Qwen) | 1,459.67 1,430–1,490 | $1.2 | $6 | 256K |
| 35 | Longcat Flash Chat ≈ | Meituan | 1,459.11 1,412–1,506 | — | — | 131K |
| 36 | Inkling ≈ | Thinking Machines Lab | 1,458.2 1,423–1,494 | $1.87 | $4.68 | 1,049K |
| 37 | Qwen3.5 397B-A17B ≈ | Alibaba (Qwen) | 1,457.54 1,442–1,473 | $0.6 | $3.6 | 262K |
| 38 | Claude Sonnet 4.5 ≈ | Anthropic | 1,457.25 1,441–1,473 | $3 | $15 | 1,000K |
| 39 | GLM 5 ≈ | Zhipu AI / Z.ai | 1,457.22 1,436–1,479 | $0.48 | $1.9 | 205K |
| 40 | Qwen3.6 Max Preview ≈ | Alibaba (Qwen) | 1,455.35 1,413–1,498 | $1.3 | $7.8 | 262K |
| 41 | DeepSeek V3.1 ≈ | DeepSeek | 1,455.24 1,418–1,492 | $0.2 | $0.7 | 164K |
| 42 | Grok 4.5 ≈ | xAI | 1,454.67 1,422–1,488 | $2 | $6 | 500K |
| 43 | GLM 5V Turbo ≈ | Zhipu AI / Z.ai | 1,454.15 1,422–1,486 | $0.7 | $3.1 | 203K |
| 44 | Gemini 3.5 Flash Lite ≈ | Google DeepMind | 1,453.97 1,412–1,496 | $0.3 | $2.5 | 1,049K |
| 45 | Kimi K2.5 Thinking TEE ≈ | Moonshot AI (Kimi) | 1,453.7 1,439–1,468 | $0.3 | $1.9 | 256K |
| 46 | Qwen3.6 Plus ≈ | Alibaba (Qwen) | 1,450.58 1,434–1,467 | $0.5 | $3 | 1,000K |
| 47 | Qwen3 235B A22B Instruct 2507 ≈ | Alibaba (Qwen) | 1,449.75 1,435–1,464 | $0.1 | $0.1 | 262K |
| 48 | Mistral Medium 3.5 ≈ | Mistral AI | 1,449.37 1,419–1,480 | $1.5 | $7.5 | 262K |
| 49 | MiMo V2 Omni ≈ | Xiaomi | 1,446.41 1,424–1,469 | $0.14 | $0.28 | 262K |
| 50 | Grok 4.1 ≈ | xAI | 1,446.22 1,430–1,463 | $2 | $10 | 200K |
| 51 | ChatGPT 4o Latest ≈ | OpenAI | 1,445.13 1,428–1,463 | $5 | $15 | 128K |
| 52 | o3 2025-04-16 ≈ | OpenAI | 1,445.04 1,422–1,468 | $2 | $8 | 200K |
| 53 | MiMo V2.5 ≈ | Xiaomi | 1,443.53 1,427–1,460 | $0.14 | $0.28 | 1,050K |
| 54 | Qwen3.5 122B-A10B ≈ | Alibaba (Qwen) | 1,442.71 1,421–1,464 | $0.4 | $3.2 | 262K |
| 55 | GPT-5.2 Chat ≈ | OpenAI | 1,442.41 1,421–1,463 | $1.75 | $14 | 128K |
| 56 | MiniMax M3 ≈ | MiniMax | 1,441.49 1,422–1,461 | $0.3 | $1.2 | 1,049K |
| 57 | Gemma 4 31B ≈ | Google DeepMind | 1,437.84 1,400–1,475 | $0.14 | $0.4 | 262K |
| 58 | 2025) ≈ | Google DeepMind | 1,437.16 1,411–1,464 | $0.3 | $2.5 | 1,049K |
| 59 | DeepSeek V3.2 Exp ≈ | DeepSeek | 1,436.53 1,394–1,479 | $0.22 | $0.33 | 164K |
| 60 | DeepSeek V4 Flash ≈ | DeepSeek | 1,433.69 1,417–1,450 | $0.14 | $0.28 | 1,049K |
| 61 | Grok 4 Fast Reasoning ≈ | xAI | 1,433.69 1,398–1,470 | $0.2 | $0.5 | 2,000K |
| 62 | Gemini 2.5 Flash ≈ | Google DeepMind | 1,431.82 1,418–1,445 | $0.3 | $2.5 | 1,049K |
| 63 | DeepSeek V3.2 ≈ | DeepSeek | 1,431.79 1,413–1,451 | $0.28 | $0.4 | 164K |
| 64 | GPT-5.6 Luna ≈ | OpenAI | 1,430.38 1,396–1,465 | $0.2 | $1.2 | 1,050K |
| 65 | GPT-5.5 Instant ≈ | OpenAI | 1,427.7 1,407–1,448 | $5 | $30 | 400K |
| 66 | Claude 4.1 Opus ≈ | Anthropic | 1,427.47 1,410–1,445 | $15 | $75 | 200K |
| 67 | GLM 4.7 ≈ | Zhipu AI / Z.ai | 1,426.59 1,390–1,463 | $0.15 | $0.8 | 205K |
| 68 | MiMo V2 Flash (Thinking) ≈ | Xiaomi | 1,426.42 1,393–1,459 | $0.1 | $0.31 | 256K |
| 69 | Grok 4 ≈ | xAI | 1,425.25 1,399–1,452 | $3 | $15 | 256K |
| 70 | GLM 4.5 ≈ | Zhipu AI / Z.ai | 1,424.96 1,391–1,459 | $0.2 | $0.8 | 131K |
| 71 | Kimi K2 Thinking Turbo ≈ | Moonshot AI (Kimi) | 1,424.58 1,408–1,442 | $1.15 | $8 | 262K |
| 72 | Gemini 3.1 Flash Lite Preview ≈ | Google DeepMind | 1,421.95 1,406–1,438 | $0.25 | $1.5 | 1,049K |
| 73 | MiniMax M2.7 ≈ | MiniMax | 1,420.92 1,405–1,437 | $0.3 | $1.2 | 205K |
| 74 | Qwen3 30B A3B Instruct 2507 ≈ | Alibaba (Qwen) | 1,420.39 1,387–1,454 | $0.05 | $0.19 | 262K |
| 75 | DeepSeek R1 0528 ≈ | DeepSeek | 1,420.34 1,381–1,460 | $0.25 | $0.25 | 164K |
| 76 | Step 3.5 Flash ≈ | StepFun | 1,419.49 1,404–1,435 | $0.1 | $0.3 | 262K |
| 77 | GPT 4.5 Preview ≈ | OpenAI | 1,418.01 1,362–1,474 | $75 | $150 | 128K |
| 78 | Qwen3 Next 80B A3B Instruct ≈ | Alibaba (Qwen) | 1,416.22 1,384–1,448 | $0.15 | $1.2 | 262K |
| 79 | Qwen3.5 35B A3B ≈ | Alibaba (Qwen) | 1,416.21 1,394–1,438 | $0.25 | $2 | 262K |
| 80 | MiMo V2 Flash ≈ | Xiaomi | 1,415.78 1,398–1,434 | $0.14 | $0.28 | 262K |
| 81 | Grok 4.1 Fast Reasoning ≈ | xAI | 1,415.7 1,398–1,433 | $0.2 | $0.5 | 2,000K |
| 82 | GPT 5 Chat ≈ | OpenAI | 1,412.24 1,382–1,442 | $1.25 | $10 | 128K |
| 83 | Qwen3.5 Flash ≈ | Alibaba (Qwen) | 1,409.81 1,395–1,425 | $0.09 | $0.36 | 1,000K |
| 84 | Nemotron 3 Super ≈ | NVIDIA | 1,408.44 1,371–1,446 | $0.2 | $0.8 | 1,000K |
| 85 | GLM 4.5 Air ≈ | Zhipu AI / Z.ai | 1,408.18 1,380–1,437 | $0.11 | $0.29 | 131K |
| 86 | Grok 4.3 ≈ | xAI | 1,407.87 1,391–1,424 | $1.25 | $2.5 | 1,000K |
| 87 | Claude Haiku 4.5 ≈ | Anthropic | 1,406.71 1,393–1,420 | $1 | $5 | 200K |
| 88 | Qwen3.5 27B ≈ | Alibaba (Qwen) | 1,405.13 1,383–1,428 | $0.3 | $2.4 | 262K |
| 89 | GPT-5.3 Chat (latest) ≈ | OpenAI | 1,398.96 1,378–1,420 | $1.75 | $14 | 128K |
| 90 | Nemotron 3 Nano 30B A3B ≈ | NVIDIA | 1,396.01 1,364–1,428 | $0.05 | $0.2 | 262K |
| 91 | Qwen3 Coder 480B A35B ≈ | Alibaba (Qwen) | 1,393.89 1,363–1,425 | $1.5 | $7.5 | 262K |
| 92 | DeepSeek Chat 0324 ≈ | DeepSeek | 1,388.26 1,362–1,414 | $0.2 | $0.6 | 164K |
| 93 | MiniMax M1 ≈ | MiniMax | 1,386.26 1,358–1,415 | $0.13 | $1.25 | 1,000K |
| 94 | Gemini 2.5 Flash Lite Preview ≈ | Google DeepMind | 1,384.87 1,355–1,415 | $0.1 | $0.4 | 1,049K |
| 95 | GPT 4.1 2025-04-14 ≈ | OpenAI | 1,381.97 1,359–1,405 | $2 | $8 | 1,048K |
| 96 | Hunyuan TurboS ≈ | Tencent (Hunyuan) | 1,381.79 1,334–1,430 | — | — | 131K |
| 97 | Trinity Large Preview ≈ | Arcee AI | 1,379.89 1,358–1,402 | — | — | 131K |
| 98 | Kimi K2 0905 Preview ≈ | Moonshot AI (Kimi) | 1,378.45 1,337–1,420 | $0.6 | $2.5 | 262K |
| 99 | Nova 2 Lite ≈ | Amazon | 1,376.95 1,337–1,417 | $0.3 | $2.5 | 1,000K |
| 100 | GPT OSS 120B ≈ | OpenAI | 1,372.17 1,343–1,402 | $0.03 | $0.14 | 131K |
The table scrolls sideways: not all columns fit.
The price is the lowest among the model's providers, base tier, without batch or discounted rates. Input and output are shown separately on purpose: for most models the output costs several times more than the input, and the final bill depends on which of the two your task has more of.
The ≈ sign marks models whose confidence interval overlaps that of the row above. Here there are 99 models — their order among themselves is not determined by the available data.
The confidence interval for half the models is wider than 45.3 points — that is about 30 adjacent rows of the table. Inside such a group the order is set not by the data but by who happened to vote this time; the data supports the difference between the top and the bottom of the list, but not the ordering of neighbouring places. Votes per model here — median 781: the fewer there are, the wider the interval.
Arena (LMArena)
— license CC-BY-4.0 · primary source
По данным Arena (LMArena)