395 моделей ИИ в одной таблице: кто лучше пишет код, рассуждает и работает агентом — и сколько это стоит. 6 из топ-10 можно подключить через IRI Gateway с оплатой в рублях.
Сравнения один на один: ChatGPT или Claude · ChatGPT или Gemini · DeepSeek или Qwen · какая нейросеть лучше
| # | Модель | Балл | Reasoning | Coding | Agent | Context | Speed | Цена API, $ за 1M | IRI Gateway, ₽ за 1M |
|---|---|---|---|---|---|---|---|---|---|
| 1 | Claude Opus 5.5в IRI GatewayAnthropic | 60.4 | 59.5 | 49.9 | 45.5 | 1M | — | $4 / $20 | 391 / 1 957 ₽ |
| 2 | Claude Sonnet 5.5в IRI GatewayAnthropic | 58.4 | 57.2 | 47.6 | 43.3 | 1M | 42 t/s | $2 / $10 | 196 / 978 ₽ |
| 3 | GPT-6 Astraв IRI GatewayOpenAI | 58.2 | 57.3 | 46.4 | 45.0 | 1.1M | — | $10 / $50 | 978 / 4 892 ₽ |
| 4 | Gemini 4 Argonв IRI GatewayGoogle | 55.1 | 53.2 | 42.5 | 41.2 | 1M | — | $2 / $10 | 196 / 930 ₽ |
| 5 | Claude Mythos PreviewAnthropic | 54.5 | 55.6 | 43.8 | 36.6 | — | — | — | — |
| 6 | Claude Opus 5Anthropic | 53.8 | 54.1 | 38.4 | 38.3 | 1M | — | $5 / $25 | — |
| 7 | GPT-6.1 Solв IRI GatewayOpenAI | 53.6 | 51.6 | 38.6 | 37.9 | 1.1M | — | $2 / $10 | 196 / 978 ₽ |
| 8 | GPT-5.6 SolOpenAI | 53.3 | 52.7 | 42.5 | 39.1 | 1.1M | — | $5 / $30 | — |
| 9 | Claude Fable 5.1в IRI GatewayAnthropic | 53.2 | 50.8 | 34.3 | 38.2 | 1M | — | $10 / $50 | 1 010 / 5 048 ₽ |
| 10 | Muse Wonder FullMeta | 53.2 | 45.7 | 42.5 | 41.0 | — | — | — | — |
| 11 | Claude Fable 5Anthropic | 53.1 | 51.2 | 42.9 | 38.6 | 1M | — | $10 / $50 | — |
| 12 | Muse Wonder FullMeta | 52.6 | 41.2 | 41.4 | 40.6 | — | — | — | — |
| 13 | Muse Spark 1.3Meta | 52.2 | 51.1 | 37.9 | 37.7 | 1M | — | $1.25 / $4.25 | — |
| 14 | Kimi K3в IRI GatewayMoonshot AI | 51.9 | 51.3 | 40.0 | 37.3 | 1M | — | $2.85 / $14.25 | 67,3 / 1 918,9 ₽ |
| 15 | DeepSeek-V4.1-Flash-1017DeepSeek | 51.2 | 48.8 | 39.3 | 38.7 | 1M | — | $0.3 / $1.2 | — |
| 16 | GLM-5.3в IRI GatewayZhipu AI | 50.9 | 50.7 | 39.4 | 37.0 | 1M | — | $1.2 / $4 | 138,3 / 434,6 ₽ |
| 17 | Qwen3.8 Maxв IRI GatewayAlibaba Cloud / Qwen Team | 50.7 | 49.7 | 37.1 | 35.7 | 1M | — | $1.65 / $4.951 | 197,5 / 592,6 ₽ |
| 18 | DeepSeek-V4-Pro-0813DeepSeek | 50.4 | 48.7 | 38.3 | 35.3 | 1M | — | $1.3 / $2.6 | — |
| 19 | DeepSeek-V4.1-Flashв IRI GatewayDeepSeek | 50.0 | 48.0 | 40.5 | 38.3 | 1M | — | $0.22 / $0.66 | 23,5 / 93,9 ₽ |
| 20 | Claude Haiku 5.5в IRI GatewayAnthropic | 49.9 | 47.5 | 39.2 | 33.8 | 1M | — | $0.1 / $0.5 | 16,4 / 81,8 ₽ |
| 21 | Claude Opus 4.8Anthropic | 49.8 | 49.9 | 38.8 | 33.9 | 1M | — | $5 / $25 | — |
| 22 | Muse WonderMeta | 49.8 | 42.4 | 39.7 | 37.6 | 1M | — | $4 / $20 | — |
| 23 | Hy4 previewTencent | 49.6 | 49.8 | 37.0 | 34.8 | — | — | — | — |
| 24 | Muse WonderMeta | 49.3 | 37.8 | 38.4 | 37.2 | — | — | — | — |
| 25 | GPT-5.6 Terraв IRI GatewayOpenAI | 49.3 | 47.9 | 38.8 | 35.8 | 1.1M | — | $2 / $12 | 196 / 1 174 ₽ |
| 26 | Gemini 3.8 Flashв IRI GatewayGoogle | 49.0 | 46.5 | 34.3 | 33.9 | 1M | — | $0.75 / $3.75 | 73,4 / 367 ₽ |
| 27 | MiMo-V2.6-Proв IRI GatewayXiaomi | 48.9 | 46.3 | 38.8 | 36.3 | 1M | — | $0.435 / $0.87 | 69 / 69 ₽ |
| 28 | GLM-5.3-Flashв IRI GatewayZhipu AI | 48.8 | 48.0 | 31.6 | 34.8 | 1M | — | $0.15 / $0.5 | 25,6 / 44 ₽ |
| 29 | MiniMax M3.1 Flashв IRI GatewayMiniMax | 48.7 | 45.9 | 37.9 | 35.7 | 1M | — | $0.15 / $0.6 | 73,3 / 73,3 ₽ |
| 30 | Muse Spark 1.1Meta | 48.3 | 49.1 | 32.0 | 32.5 | 1M | — | $1.25 / $4.25 | — |
| 31 | Qwen3.8-Flash-NextAlibaba Cloud / Qwen Team | 48.1 | 48.5 | 33.2 | 32.8 | — | — | — | — |
| 32 | Qwen3.8 FlashAlibaba Cloud / Qwen Team | 48.1 | 48.4 | 33.1 | 32.8 | 1M | — | $0.15 / $0.47 | — |
| 33 | GPT-6 SolOpenAI | 47.8 | 45.3 | 33.0 | 34.9 | 1.1M | — | $2 / $10 | — |
| 34 | Gemini 3.7 FlashGoogle | 47.6 | 45.4 | 31.7 | 31.6 | 1M | — | $0.75 / $3.75 | — |
| 35 | GPT-5.5OpenAI | 47.2 | 47.2 | 35.6 | 31.3 | 1.1M | — | $5 / $30 | — |
| 36 | Claude Sonnet 5Anthropic | 47.2 | 46.9 | 34.1 | 29.9 | 1M | 42 t/s | $2 / $10 | — |
| 37 | Grok 4.7в IRI GatewayxAI | 46.7 | 43.0 | 35.0 | 33.9 | 500K | — | $2 / $6 | 259,2 / 777,6 ₽ |
| 38 | GPT-5.5 ProOpenAI | 45.9 | 45.9 | — | 31.1 | — | — | — | — |
| 39 | Mistral Large 4Mistral AI | 45.8 | 43.9 | 35.0 | 33.8 | 1M | — | $0.68 / $2.09 | — |
| 40 | Seed 2.1 ProByteDance | 45.7 | 47.2 | 32.0 | 31.0 | — | — | — | — |
| 41 | Grok 4.5xAI | 45.5 | 44.8 | 32.0 | 30.2 | 500K | 80 t/s | $2 / $6 | — |
| 42 | Muse Spark OpenMeta | 45.3 | 45.7 | 33.5 | 30.8 | 1M | — | $1.25 / $4.25 | — |
| 43 | Qwen3.7 MaxAlibaba Cloud / Qwen Team | 45.1 | 45.9 | 34.9 | 28.4 | 1M | — | $1.25 / $3.75 | — |
| 44 | GLM-5.2Zhipu AI | 45.0 | 44.7 | 33.4 | 28.8 | 1M | — | $0.75 / $2.4 | — |
| 45 | MiMo-V2.6-FlashXiaomi | 44.8 | 43.6 | 34.0 | 32.0 | 1M | — | $0.14 / $0.28 | — |
| 46 | Beam 1.0Reflection AI | 44.6 | 45.5 | 33.7 | 28.8 | 1M | — | $0.4 / $1.2 | — |
| 47 | Grok 4.6xAI | 44.5 | 47.0 | 32.5 | 34.0 | 500K | — | $2 / $6 | — |
| 48 | Claude Opus 4.6Anthropic | 44.5 | 45.3 | 31.3 | 27.4 | 1M | 42 t/s | $5 / $25 | — |
| 49 | Qwen3.8-27BAlibaba Cloud / Qwen Team | 44.4 | 44.1 | 29.5 | 29.4 | 262K | — | $0.4 / $3 | — |
| 50 | Claude Mythos 5.1Anthropic | 44.3 | 43.3 | 36.1 | 34.1 | — | — | — | — |
| 51 | GPT-5.6 LunaOpenAI | 44.2 | 43.3 | 33.8 | 30.0 | 1.1M | — | $0.2 / $1.2 | — |
| 52 | DeepSeek-V4-Flash-Vision-ExpDeepSeek | 44.2 | 40.8 | 33.0 | 30.4 | 1M | — | $0.44 / $1.32 | — |
| 53 | Claude Opus 4.7Anthropic | 43.5 | 45.4 | 34.9 | 30.1 | 1M | 42 t/s | $5 / $25 | — |
| 54 | Kimi K2.6Moonshot AI | 43.4 | 43.8 | 32.9 | 26.7 | 262K | — | $0.75 / $3.5 | — |
| 55 | GPT-6 Lunaв IRI GatewayOpenAI | 43.1 | 39.9 | 28.6 | 30.1 | 1.1M | — | $0.1 / $0.5 | 9,8 / 48,9 ₽ |
| 56 | Gemini 3.1 ProGoogle | 43.0 | 44.9 | 30.0 | 24.0 | 1M | — | $2 / $12 | — |
| 57 | DeepSeek-V4-Flash-0731DeepSeek | 43.0 | 41.1 | 30.8 | 28.9 | 1M | — | $0.06 / $0.18 | — |
| 58 | GPT-5.2 ProOpenAI | 43.0 | 41.6 | — | 18.2 | — | — | — | — |
| 59 | Ling 3.0 Flash FinInclusionAI | 42.9 | 44.7 | — | 28.6 | 262K | — | $0.06 / $0.18 | — |
| 60 | DeepSeek-V4-Pro-MaxDeepSeek | 42.8 | 43.1 | 31.2 | 23.5 | 1M | — | $1.3 / $2.6 | — |
| 61 | Hy3Tencent | 42.5 | 42.7 | 31.9 | 26.4 | 262K | — | $0.14 / $0.58 | — |
| 62 | Seed 2.1 TurboByteDance | 42.5 | 44.2 | 29.4 | 28.1 | — | — | — | — |
| 63 | Gemini 3.6 FlashGoogle | 42.4 | 39.6 | 26.7 | 26.3 | 1M | — | $1.5 / $7.5 | — |
| 64 | Qwen3.7-PlusAlibaba Cloud / Qwen Team | 42.3 | 42.9 | 29.9 | 23.8 | — | — | — | — |
| 65 | Ember-1Fireworks AI | 42.2 | 38.7 | 35.9 | 28.6 | 1M | — | $3 / $15 | — |
| 66 | Gemini 3.5 FlashGoogle | 42.0 | 43.2 | 29.5 | 28.3 | 1M | — | $1.5 / $9 | — |
| 67 | Sakana NamazuSakana AI | 41.7 | 41.5 | — | — | 256K | — | $0.95 / $4 | — |
| 68 | GPT-5.4OpenAI | 41.6 | 42.2 | 29.5 | 26.1 | 1M | 50 t/s | $2.5 / $15 | — |
| 69 | Muse SparkMeta | 41.6 | 40.5 | 22.9 | 15.2 | — | — | — | — |
| 70 | GPT-5.2OpenAI | 41.0 | 40.5 | 23.3 | 14.4 | 400K | 100 t/s | $1.75 / $14 | — |
| 71 | MiniMax M3в IRI GatewayMiniMax | 40.9 | 40.8 | 30.7 | 25.7 | 1M | — | $0.28 / $1.1 | 48 / 191,9 ₽ |
| 72 | Grok-4 HeavyxAI | 40.6 | 38.9 | 17.5 | — | — | — | — | — |
| 73 | Laguna S 2.1Poolside | 39.7 | 39.7 | 29.1 | 23.3 | 1M | — | $0.1 / $0.2 | — |
| 74 | Gemini 3.8 Flash CyberGoogle | 39.3 | — | 31.8 | 28.6 | — | — | — | — |
| 75 | Seed 2.0 ProByteDance | 39.1 | 38.7 | 21.9 | 17.3 | 256K | — | $0.5 / $3 | — |
| 76 | Qwen3.6 PlusAlibaba Cloud / Qwen Team | 39.0 | 40.1 | 27.2 | 20.3 | 1M | — | $0.5 / $3 | — |
| 77 | DeepSeek-V4-Flash-MaxDeepSeek | 38.9 | 39.6 | 28.5 | 18.1 | 1M | — | $0.09 / $0.18 | — |
| 78 | GLM-5.1Zhipu AI | 38.8 | 38.7 | 29.3 | 23.7 | 202K | — | $1.05 / $3.5 | — |
| 79 | Kimi K2.7 Codeв IRI GatewayMoonshot AI | 38.7 | 33.8 | 28.7 | 25.9 | 262K | — | $0.68 / $3.4 | 104 / 298 ₽ |
| 80 | Kimi K2.5Moonshot AI | 38.5 | 38.4 | 24.5 | 18.1 | — | — | — | — |
| 81 | Gemini 3 ProGoogle | 38.3 | 38.4 | 24.3 | 15.2 | — | — | — | — |
| 82 | Qwen3.5-397B-A17BAlibaba Cloud / Qwen Team | 38.3 | 38.2 | 22.8 | 19.8 | 262K | — | $0.45 / $3 | — |
| 83 | Inkling-SmallThinking Machines Lab | 38.2 | 37.6 | 28.2 | 22.5 | 524K | — | $0.3 / $1.2 | — |
| 84 | Claude Opus 4.5Anthropic | 37.9 | 39.8 | 27.6 | 17.4 | — | — | — | — |
| 85 | Qwen3.8 Omni FlashAlibaba Cloud / Qwen Team | 37.8 | 32.1 | — | 21.2 | — | — | — | — |
| 86 | InklingThinking Machines Lab | 37.1 | 36.8 | 26.0 | 19.9 | 524K | — | $0.95 / $4.05 | — |
| 87 | Gemini 3.5 Flash CyberGoogle | 37.1 | — | 29.4 | 26.0 | — | — | — | — |
| 88 | Muse Spark 1.2Meta | 37.0 | 33.0 | 27.0 | 24.0 | 1M | — | $1.25 / $4.25 | — |
| 89 | GLM-5Zhipu AI | 36.8 | 37.2 | 25.6 | 19.4 | 200K | 30 t/s | $1 / $3.2 | — |
| 90 | Step-3.5-FlashStepFun | 36.8 | 36.9 | 20.4 | 15.2 | 65K | 150 t/s | $0.1 / $0.4 | — |
| 91 | MiniMax M2.5MiniMax | 36.6 | 36.7 | 26.5 | 17.1 | 1M | 100 t/s | $0.3 / $1.2 | — |
| 92 | Claude Sonnet 4.6Anthropic | 36.6 | 37.9 | 24.6 | 23.6 | 1M | 42 t/s | $3 / $15 | — |
| 93 | Gemini 3 FlashGoogle | 36.5 | 37.5 | 22.0 | 17.5 | 1M | — | $0.5 / $3 | — |
| 94 | GPT-5.1 ThinkingOpenAI | 36.4 | 36.2 | 21.5 | — | — | — | — | — |
| 95 | GPT-5.1 InstantOpenAI | 36.3 | 36.1 | 21.3 | — | 400K | 100 t/s | $1.25 / $10 | — |
| 96 | GPT-5.3 CodexOpenAI | 36.3 | 37.0 | 29.5 | 22.7 | 400K | 50 t/s | $1.75 / $14 | — |
| 97 | GPT-5.1OpenAI | 36.2 | 35.9 | 21.1 | — | 400K | 100 t/s | $1.25 / $10 | — |
| 98 | Mercury 3Inception | 36.1 | 34.8 | 21.5 | 19.4 | 1M | 1068 t/s | $0.1 / $0.35 | — |
| 99 | DeepSeek-V4-Flash-0423DeepSeek | 35.9 | 36.9 | 26.6 | 15.6 | 1M | — | $0.09 / $0.18 | — |
| 100 | Ling 3.0 FlashInclusionAI | 35.9 | 35.2 | 24.4 | 19.0 | 131K | — | $0.06 / $0.18 | — |
| 101 | Nemotron 3 Ultra (550B A55B)NVIDIA | 35.8 | 35.1 | 19.6 | 12.0 | 262K | — | $0.5 / $2.2 | — |
| 102 | MiMo-V2.5Xiaomi | 35.8 | 35.4 | 28.6 | 20.8 | 1M | — | $0.168 / $0.336 | — |
| 103 | MiMo-V2-ProXiaomi | 35.5 | 37.7 | 24.9 | 17.1 | — | — | — | — |
| 104 | Kimi K2-Thinking-0905Moonshot AI | 35.5 | 35.8 | 20.1 | 15.6 | — | — | — | — |
| 105 | MiniMax M2.7MiniMax | 35.4 | 34.5 | 27.1 | 15.6 | 204K | — | $0.3 / $1.2 | — |
| 106 | Muse Glimmer-30BMeta | 35.2 | 34.7 | 22.5 | 17.3 | 131K | — | $0.3 / $1.2 | — |
| 107 | Qwen3.6-27BAlibaba Cloud / Qwen Team | 35.2 | 35.9 | 25.3 | 15.9 | 262K | — | $0.32 / $3.2 | — |
| 108 | GPT-5.1 HighOpenAI | 35.0 | 34.5 | — | — | — | — | — | — |
| 109 | LongCat-Flash-Thinking-2601Meituan | 34.9 | 35.0 | 18.6 | 9.2 | — | — | — | — |
| 110 | Nemotron 3.5 Super VL (120B A12B)NVIDIA | 34.8 | 36.0 | 20.9 | 16.4 | 262K | — | $0.2 / $1 | — |
| 111 | Qwen3.5-122B-A10BAlibaba Cloud / Qwen Team | 34.7 | 35.1 | 20.2 | 15.4 | 262K | — | $0.29 / $2.4 | — |
| 112 | Solar Pro 4Upstage | 34.7 | 34.4 | 20.5 | 11.5 | 524K | — | $0.3 / $1.2 | — |
| 113 | GPT-5.2 CodexOpenAI | 34.1 | 34.5 | 24.8 | 19.0 | 400K | 50 t/s | $1.75 / $14 | — |
| 114 | GLM-4.7Zhipu AI | 34.0 | 33.9 | 17.3 | 9.2 | 202K | — | $0.4 / $1.75 | — |
| 115 | Seed 1.8ByteDance | 34.0 | 33.3 | 19.5 | 14.6 | 256K | — | $0.25 / $2 | — |
| 116 | GPT-5 HighOpenAI | 33.7 | 32.6 | — | — | — | — | — | — |
| 117 | Qwen3.5-27BAlibaba Cloud / Qwen Team | 33.5 | 34.3 | 16.7 | 14.6 | 262K | — | $0.26 / $2.6 | — |
| 118 | DeepSeek-V3.2-SpecialeDeepSeek | 33.5 | 32.1 | 18.5 | 10.8 | — | — | — | — |
| 119 | Grok-4xAI | 33.5 | 32.7 | 17.1 | — | — | — | — | — |
| 120 | GPT-5OpenAI | 33.4 | 33.8 | 26.2 | 8.7 | — | — | — | — |
| 121 | ERNIE 5.0Baidu | 33.2 | 32.8 | — | — | — | — | — | — |
| 122 | DeepSeek-V3.2DeepSeek | 32.9 | 32.9 | 22.8 | 11.9 | 163K | — | $0.26 / $0.38 | — |
| 123 | Gemma 4 31BGoogle | 32.7 | 33.2 | — | 13.6 | 262K | — | $0.09 / $0.34 | — |
| 124 | MAI-Thinking-1Microsoft | 32.6 | 33.5 | 19.2 | 12.8 | — | — | — | — |
| 125 | Qwen3 Max ThinkingAlibaba Cloud / Qwen Team | 32.6 | 33.7 | 16.0 | 9.9 | 256K | — | $1.2 / $6 | — |
| 126 | Solar Mini 4Upstage | 32.4 | 32.3 | 17.4 | 2.6 | 524K | — | $0.1 / $0.4 | — |
| 127 | DeepSeek-V3.2 (Thinking)DeepSeek | 32.2 | 32.2 | 22.7 | 10.9 | — | — | — | — |
| 128 | Qwen3.6-35B-A3BAlibaba Cloud / Qwen Team | 32.0 | 32.6 | 19.0 | 11.7 | 262K | — | $0.1 / $0.95 | — |
| 129 | GPT-5.5 InstantOpenAI | 31.8 | 30.7 | — | — | 400K | 100 t/s | $5 / $30 | — |
| 130 | Claude Sonnet 4.5Anthropic | 31.7 | 30.9 | 20.8 | 17.0 | 200K | 42 t/s | $3 / $15 | — |
| 131 | GPT-RosalindOpenAI | 31.6 | 31.1 | — | 21.5 | — | — | $5 / $25 | — |
| 132 | Seed 2.0 LiteByteDance | 31.5 | 31.1 | 17.9 | — | — | — | — | — |
| 133 | MiniMax M2.1MiniMax | 31.3 | 31.7 | 21.6 | 17.6 | 1M | 100 t/s | $0.3 / $1.2 | — |
| 134 | Grok 4 FastxAI | 31.2 | 30.2 | 17.9 | 2.7 | — | — | — | — |
| 135 | GPT-5.1 MediumOpenAI | 30.9 | 30.8 | — | — | 400K | 90 t/s | $1.25 / $10 | — |
| 136 | Grok-3 MinixAI | 30.9 | 30.5 | 18.5 | — | — | — | — | — |
| 137 | o3OpenAI | 30.6 | 30.4 | 14.6 | 3.6 | — | — | — | — |
| 138 | Qwen3.5-35B-A3BAlibaba Cloud / Qwen Team | 30.5 | 31.6 | 11.4 | 12.5 | 262K | — | $0.14 / $1 | — |
| 139 | Gemini 3.1 Flash-LiteGoogle | 30.5 | 30.9 | — | 10.4 | 1M | — | $0.25 / $1.5 | — |
| 140 | Nova 2 ProAmazon | 30.5 | 30.3 | 19.0 | 11.9 | — | — | — | — |
| 141 | MiMo-V2-FlashXiaomi | 30.3 | 30.5 | 17.3 | 7.9 | — | — | — | — |
| 142 | GLM-5V-TurboZhipu AI | 30.0 | 7.7 | 22.5 | 18.4 | — | — | — | — |
| 143 | Grok-3xAI | 29.9 | 29.7 | 17.4 | — | 128K | 100 t/s | $3 / $15 | — |
| 144 | Step 3.7 FlashStepFun | 29.7 | 29.8 | 19.0 | 13.1 | 262K | — | $0.2 / $1.15 | — |
| 145 | GPT-5.4 miniOpenAI | 29.7 | 31.6 | 21.5 | 19.0 | 400K | — | $0.75 / $4.5 | — |
| 146 | Nemotron 3.5 Bolt (74B A9B 24GB)NVIDIA | 29.6 | 28.9 | 14.4 | 11.9 | — | — | — | — |
| 147 | Claude Opus 4.1Anthropic | 29.5 | 29.6 | 21.3 | 12.7 | — | — | — | — |
| 148 | Gemini 3.5 Flash-LiteGoogle | 29.5 | 25.7 | 15.5 | 15.4 | 1M | — | $0.3 / $2.5 | — |
| 149 | Gemini 2.5 Pro Preview 06-05Google | 29.5 | 27.9 | 16.5 | — | — | — | — | — |
| 150 | GPT-5.1 CodexOpenAI | 29.4 | 29.5 | 20.9 | 10.4 | — | — | — | — |
| 151 | GPT-5 MediumOpenAI | 29.3 | 28.2 | — | — | — | — | — | — |
| 152 | Gemma 4 26B-A4BGoogle | 29.2 | 28.8 | — | 12.6 | 262K | — | $0.07 / $0.34 | — |
| 153 | GPT-5.1 Codex HighOpenAI | 29.1 | 28.9 | — | — | — | — | — | — |
| 154 | GPT-5.3 ChatOpenAI | 29.0 | — | — | — | 128K | 100 t/s | $1.75 / $14 | — |
| 155 | MAI-Code-1-FlashMicrosoft | 28.6 | 28.2 | 19.0 | 11.1 | — | — | — | — |
| 156 | GLM-4.6Zhipu AI | 28.5 | 28.5 | 14.5 | 8.9 | 202K | — | $0.5 / $2 | — |
| 157 | LongCat-Flash-ThinkingMeituan | 28.2 | 28.4 | 13.6 | 12.9 | — | — | — | — |
| 158 | GPT OSS 120BOpenAI | 28.1 | 22.6 | — | — | 131K | — | $0.037 / $0.17 | — |
| 159 | Seed 2.0 MiniByteDance | 27.8 | 28.7 | 1.5 | 1.1 | 256K | — | $0.1 / $0.4 | — |
| 160 | DeepSeek-V3.2-ExpDeepSeek | 27.7 | 27.6 | 17.1 | 5.6 | — | — | — | — |
| 161 | Qwen3-235B-A22B-Thinking-2507Alibaba Cloud / Qwen Team | 27.6 | 28.0 | — | 11.3 | — | — | — | — |
| 162 | MAI-Code-1.1-FlashMicrosoft | 27.4 | 27.3 | 18.2 | 10.3 | 256K | — | $0.2 / $1.2 | — |
| 163 | Claude Opus 4Anthropic | 27.4 | 27.4 | 17.2 | 8.0 | — | — | — | — |
| 164 | Gemini 2.5 ProGoogle | 27.4 | 26.8 | 12.5 | — | 1M | 85 t/s | $1.25 / $10 | — |
| 165 | GPT OSS 20B HighOpenAI | 27.3 | 27.4 | — | — | — | — | — | — |
| 166 | GLM-4.5Zhipu AI | 27.3 | 26.8 | 15.6 | 9.6 | — | — | — | — |
| 167 | o4-miniOpenAI | 27.0 | 27.0 | 11.6 | 6.6 | — | — | — | — |
| 168 | MiMo-V2-OmniXiaomi | 26.9 | 26.6 | 19.1 | 8.0 | — | — | — | — |
| 169 | GPT-5 miniOpenAI | 26.9 | 26.2 | — | — | 400K | 200 t/s | $0.25 / $2 | — |
| 170 | GPT-5 CodexOpenAI | 26.5 | 26.4 | 19.3 | — | — | — | — | — |
| 171 | MiniMax M2MiniMax | 26.4 | 26.1 | 17.6 | 9.1 | 1M | — | $0.3 / $1.2 | — |
| 172 | Mistral Medium 3.5Mistral AI | 26.4 | 26.9 | 23.6 | 8.2 | 256K | — | $1.5 / $7.5 | — |
| 173 | Qwen3 VL 235B A22B ThinkingAlibaba Cloud / Qwen Team | 26.3 | 26.5 | — | 13.0 | — | — | — | — |
| 174 | GPT-5.4 nanoOpenAI | 26.3 | 27.2 | 15.7 | 11.2 | 400K | — | $0.2 / $1.25 | — |
| 175 | Nemotron 3 Super (120B A12B)NVIDIA | 26.2 | 24.5 | 12.7 | -0.8 | 262K | — | $0.085 / $0.4 | — |
| 176 | K-EXAONE-236B-A23BLG AI Research | 25.8 | 26.8 | — | 4.0 | — | — | — | — |
| 177 | EXAONE 4.5 33BLG AI Research | 25.5 | 26.1 | — | — | — | — | — | — |
| 178 | Nova 2 OmniAmazon | 25.3 | 27.6 | — | 4.7 | — | — | — | — |
| 179 | MiMo-V2.5-ProXiaomi | 25.3 | 23.7 | 32.2 | 23.6 | 1M | — | $0.435 / $0.87 | — |
| 180 | Claude Haiku 4.5Anthropic | 25.2 | 24.9 | 19.4 | 13.2 | 200K | — | $1 / $5 | — |
| 181 | Sarvam-105BSarvam AI | 25.2 | 25.7 | -2.2 | 5.1 | — | — | — | — |
| 182 | Step3-VL-10BStepFun | 25.2 | 25.2 | — | — | — | — | — | — |
| 183 | GPT OSS 120B HighOpenAI | 25.0 | 25.2 | — | 3.0 | — | — | — | — |
| 184 | Nova 2 LiteAmazon | 25.0 | 26.6 | 12.0 | 4.6 | 1M | — | $0.3 / $2.5 | — |
| 185 | ChatGPT-4o LatestOpenAI | 24.4 | 23.6 | — | — | — | — | — | — |
| 186 | Qwen3 VL 235B A22B InstructAlibaba Cloud / Qwen Team | 24.3 | 22.1 | — | 12.5 | 262K | — | $0.2 / $0.88 | — |
| 187 | Qwen3.5-9BAlibaba Cloud / Qwen Team | 24.2 | 23.9 | — | 8.4 | 262K | — | $0.1 / $0.15 | — |
| 188 | Laguna XS 2.1Poolside | 23.9 | 24.0 | 14.2 | 2.6 | 262K | — | $0.1 / $0.2 | — |
| 189 | GLM-4.5-AirZhipu AI | 23.9 | 23.3 | 10.8 | 5.7 | — | — | — | — |
| 190 | Claude Sonnet 4Anthropic | 23.8 | 23.9 | 16.5 | 5.5 | — | — | — | — |
| 191 | Qwen3-235B-A22B-Instruct-2507Alibaba Cloud / Qwen Team | 23.7 | 23.5 | 3.9 | 9.8 | 262K | — | $0.09 / $0.55 | — |
| 192 | DeepSeek-R1-0528DeepSeek | 23.6 | 23.2 | 7.4 | -12.2 | 163K | — | $0.5 / $2.15 | — |
| 193 | Command A+Cohere | 23.6 | 21.0 | 6.3 | — | — | — | — | — |
| 194 | Claude 3.7 SonnetAnthropic | 23.5 | 22.6 | 13.3 | 4.7 | — | — | — | — |
| 195 | Grok-4.20 Beta ReasoningxAI | 23.4 | 15.0 | — | 2.8 | — | — | — | — |
| 196 | Qwen3-Next-80B-A3B-ThinkingAlibaba Cloud / Qwen Team | 23.4 | 24.1 | — | 12.1 | — | — | — | — |
| 197 | IBM Granite 4.2 30BIBM | 23.3 | 21.0 | 8.3 | 7.4 | 131K | — | $0.16 / $0.65 | — |
| 198 | Qwen3 VL 32B ThinkingAlibaba Cloud / Qwen Team | 23.2 | 24.3 | — | 12.8 | — | — | — | — |
| 199 | Mercury 2Inception | 23.1 | 22.2 | 11.9 | — | 128K | 1009 t/s | $0.25 / $0.75 | — |
| 200 | U2Unisound | 23.0 | 19.5 | 20.6 | 4.5 | — | — | — | — |
| 201 | GLM-4.7-FlashZhipu AI | 23.0 | 23.0 | 3.7 | 4.1 | — | — | — | — |
| 202 | Grok 4.3xAI | 22.0 | 22.5 | — | 10.1 | 1M | — | $1.25 / $2.5 | — |
| 203 | Nemotron Labs Dory 1.0 (4B)NVIDIA | 21.8 | 19.6 | 2.8 | -1.8 | — | — | — | — |
| 204 | Nemotron 3.5 Lightning (30B A3B)NVIDIA | 21.8 | 19.8 | 7.1 | 5.1 | 262K | — | $0.08 / $0.2 | — |
| 205 | Gemma 4 12BGoogle | 21.6 | 22.5 | — | — | — | — | — | — |
| 206 | DeepSeek-V3.1DeepSeek | 21.6 | 21.9 | 12.2 | 0.0 | 163K | — | $0.25 / $0.95 | — |
| 207 | North Mini Code 1.0Cohere | 21.5 | 21.5 | 10.6 | 0.8 | — | — | — | — |
| 208 | Kimi K2 InstructMoonshot AI | 21.4 | 21.5 | 12.0 | -1.2 | — | — | — | — |
| 209 | Gemini 2.5 FlashGoogle | 21.3 | 21.0 | 8.1 | — | 1M | — | $0.3 / $2.5 | — |
| 210 | Qwen3 MaxAlibaba Cloud / Qwen Team | 21.3 | 21.5 | 11.1 | 5.0 | 256K | — | $1.2 / $6 | — |
| 211 | Kimi K2 0905Moonshot AI | 21.2 | 21.5 | 19.3 | — | — | — | — | — |
| 212 | Kimi K2-Instruct-0905Moonshot AI | 21.1 | 21.4 | 9.4 | -4.3 | — | — | — | — |
| 213 | o3-miniOpenAI | 21.0 | 20.1 | 4.7 | — | — | — | — | — |
| 214 | MiniMax M1 80KMiniMax | 20.9 | 21.1 | 9.1 | — | — | — | — | — |
| 215 | Qwen3-Coder 480B A35B InstructAlibaba Cloud / Qwen Team | 20.6 | 20.7 | 11.9 | 5.2 | 262K | — | $0.3 / $1 | — |
| 216 | Nemotron 3 Nano (30B A3B)NVIDIA | 20.5 | 19.1 | -0.3 | -6.9 | 262K | — | $0.05 / $0.2 | — |
| 217 | o1OpenAI | 20.5 | 20.9 | 4.2 | — | — | — | — | — |
| 218 | Grok Code Fast 1xAI | 20.4 | 20.3 | 12.6 | — | 256K | 76 t/s | $0.2 / $1.5 | — |
| 219 | Ministral 3 (14B Reasoning 2512)Mistral AI | 20.1 | 20.1 | 11.0 | — | — | — | — | — |
| 220 | MiniMax M1 40KMiniMax | 20.0 | 20.2 | 7.7 | — | — | — | — | — |
| 221 | Qwen3 VL 32B InstructAlibaba Cloud / Qwen Team | 19.8 | 18.7 | — | 9.4 | — | — | — | — |
| 222 | Qwen3-Next-80B-A3B-InstructAlibaba Cloud / Qwen Team | 19.7 | 19.4 | 1.6 | 8.2 | 262K | — | $0.09 / $1.1 | — |
| 223 | GPT-4.5OpenAI | 19.7 | 19.9 | 5.5 | — | — | — | — | — |
| 224 | IBM Granite 4.2 8BIBM | 19.3 | 17.1 | 2.8 | 2.7 | 131K | — | $0.06 / $0.25 | — |
| 225 | LongCat-Flash-ChatMeituan | 19.3 | 19.3 | 12.2 | 8.9 | — | — | — | — |
| 226 | Grok-4.1xAI | 19.3 | 20.3 | — | — | — | — | — | — |
| 227 | o1-proOpenAI | 19.2 | 19.6 | — | — | — | — | — | — |
| 228 | Qwen3.5-4BAlibaba Cloud / Qwen Team | 19.0 | 19.1 | — | 3.6 | — | — | — | — |
| 229 | Sarvam-30BSarvam AI | 19.0 | 18.8 | 5.6 | -1.7 | — | — | — | — |
| 230 | DiffusionGemma 26B-A4BGoogle | 19.0 | 18.7 | — | 0.4 | — | — | — | — |
| 231 | GPT-5 nanoOpenAI | 18.9 | 18.2 | — | — | — | — | — | — |
| 232 | Llama 3.1 Nemotron Ultra 253B v1NVIDIA | 18.7 | 18.4 | 12.2 | — | — | — | — | — |
| 233 | Claude 3.5 SonnetAnthropic | 18.6 | 19.6 | 11.1 | — | — | — | — | — |
| 234 | LFM2.5-2.6BLiquid AI | 18.4 | 14.1 | 16.0 | 6.7 | — | — | — | — |
| 235 | Mistral Small 4Mistral AI | 18.3 | 19.2 | 10.5 | — | 256K | — | $0.15 / $0.6 | — |
| 236 | LongCat-Flash-LiteMeituan | 18.1 | 18.4 | 6.6 | 3.8 | 256K | 500 t/s | $0.1 / $0.4 | — |
| 237 | Qwen3 32BAlibaba Cloud / Qwen Team | 18.0 | 18.3 | 8.7 | — | 128K | — | $0.08 / $0.28 | — |
| 238 | Nemotron Nano 9B v2NVIDIA | 17.9 | 16.7 | 14.3 | — | — | — | — | — |
| 239 | Qwen3 VL 30B A3B ThinkingAlibaba Cloud / Qwen Team | 17.8 | 19.1 | — | 7.0 | — | — | — | — |
| 240 | Qwen3 14BAlibaba Cloud / Qwen Team | 17.6 | 19.1 | — | 9.0 | 40K | — | $0.12 / $0.24 | — |
| 241 | GPT OSS 20BOpenAI | 17.5 | 15.1 | — | — | 131K | — | $0.03 / $0.14 | — |
| 242 | GPT-4.1OpenAI | 17.5 | 17.9 | 4.6 | — | 1M | 100 t/s | $2 / $8 | — |
| 243 | Qwen3 30B A3BAlibaba Cloud / Qwen Team | 16.8 | 17.2 | 9.7 | — | 128K | 89 t/s | $0.1 / $0.44 | — |
| 244 | Kimi-k1.5Moonshot AI | 16.7 | 15.9 | — | — | — | — | — | — |
| 245 | Gemini 2.0 Flash ThinkingGoogle | 16.3 | 16.6 | — | — | — | — | — | — |
| 246 | o1-previewOpenAI | 16.1 | 16.3 | -3.7 | — | — | — | — | — |
| 247 | Gemini 2.0 FlashGoogle | 16.0 | 15.4 | 1.5 | — | — | — | — | — |
| 248 | Ministral 3 (8B Reasoning 2512)Mistral AI | 16.0 | 16.1 | 8.8 | — | — | — | — | — |
| 249 | Qwen3 VL 30B A3B InstructAlibaba Cloud / Qwen Team | 15.9 | 14.5 | — | 4.3 | 262K | — | $0.15 / $0.6 | — |
| 250 | Qwen3 VL 8B ThinkingAlibaba Cloud / Qwen Team | 15.8 | 16.7 | — | 4.2 | — | — | — | — |
| 251 | Grok-4.1 ThinkingxAI | 15.7 | — | 1.1 | 1.3 | — | — | — | — |
| 252 | DeepSeek R1 ZeroDeepSeek | 15.6 | 15.8 | 4.1 | — | — | — | — | — |
| 253 | Claude 3.5 SonnetAnthropic | 15.4 | 15.5 | 12.8 | — | — | — | — | — |
| 254 | DeepSeek-V3DeepSeek | 15.1 | 14.2 | 5.9 | — | 163K | — | $0.32 / $0.89 | — |
| 255 | Qwen3 235B A22BAlibaba Cloud / Qwen Team | 15.1 | 15.8 | 12.1 | — | — | — | — | — |
| 256 | IBM Granite 4.2 3BIBM | 15.0 | 11.3 | -7.9 | 1.6 | 131K | — | $0.03 / $0.12 | — |
| 257 | Phi 4 Reasoning PlusMicrosoft | 14.5 | 14.9 | 5.5 | — | — | — | — | — |
| 258 | QwQ-32BAlibaba Cloud / Qwen Team | 14.3 | 15.1 | 10.1 | — | — | — | — | — |
| 259 | Llama 3.1 405B InstructMeta | 14.2 | 13.3 | 11.4 | — | — | — | — | — |
| 260 | Llama 4 MaverickMeta | 14.2 | 13.8 | 2.9 | — | 1M | — | $0.2 / $0.8 | — |
| 261 | DeepSeek R1 Distill Llama 70BDeepSeek | 13.9 | 14.2 | 8.4 | — | — | — | — | — |
| 262 | MiniCPM-SALAOpenBMB | 13.9 | 15.2 | 20.9 | — | — | — | — | — |
| 263 | Qwen2-VL-72B-InstructAlibaba Cloud / Qwen Team | 13.9 | 11.2 | — | — | — | — | — | — |
| 264 | GPT-4oOpenAI | 13.6 | 15.4 | -1.0 | — | 128K | 132 t/s | $2.5 / $10 | — |
| 265 | GPT-4.1 miniOpenAI | 13.5 | 13.4 | -3.7 | — | 1M | 150 t/s | $0.4 / $1.6 | — |
| 266 | Qwen3 VL 8B InstructAlibaba Cloud / Qwen Team | 13.4 | 10.8 | — | 7.0 | — | — | — | — |
| 267 | Gemma 4 E4BGoogle | 13.3 | 13.2 | — | 1.8 | 131K | — | $0.02 / $0.1 | — |
| 268 | Llama 3.3 70B InstructMeta | 13.3 | 11.2 | 9.2 | — | 131K | — | $0.1 / $0.32 | — |
| 269 | Devstral MediumMistral AI | 13.3 | 13.2 | 5.1 | — | — | — | — | — |
| 270 | Kimi K2 BaseMoonshot AI | 13.0 | 12.1 | 10.5 | — | — | — | — | — |
| 271 | DeepSeek-V3 0324DeepSeek | 12.8 | 13.1 | 3.8 | — | 163K | — | $0.24 / $0.9 | — |
| 272 | Llama-3.3 Nemotron Super 49B v1NVIDIA | 12.6 | 12.8 | — | — | — | — | — | — |
| 273 | DeepSeek R1 Distill Qwen 32BDeepSeek | 12.5 | 12.8 | 8.0 | — | — | — | — | — |
| 274 | Qwen3 VL 4B ThinkingAlibaba Cloud / Qwen Team | 12.5 | 13.6 | — | 6.5 | — | — | — | — |
| 275 | Magistral MediumMistral AI | 12.5 | 12.7 | 3.3 | — | — | — | — | — |
| 276 | Qwen2.5 VL 72B InstructAlibaba Cloud / Qwen Team | 12.2 | 12.3 | — | 4.5 | — | — | — | — |
| 277 | Gemini 2.0 Flash-LiteGoogle | 11.9 | 12.0 | — | — | — | — | — | — |
| 278 | Nova ProAmazon | 11.8 | 11.5 | 10.0 | 10.2 | — | — | — | — |
| 279 | Qwen2.5 72B InstructAlibaba Cloud / Qwen Team | 11.6 | 11.7 | 9.0 | — | 32K | — | $0.36 / $0.4 | — |
| 280 | Phi 4 ReasoningMicrosoft | 11.5 | 12.1 | 6.2 | — | — | — | — | — |
| 281 | Gemini 1.5 ProGoogle | 11.5 | 11.3 | 4.2 | — | — | — | — | — |
| 282 | Pixtral LargeMistral AI | 11.3 | 12.7 | — | — | — | — | — | — |
| 283 | GPT-4oOpenAI | 11.0 | 11.7 | 11.6 | — | 128K | 100 t/s | $2.5 / $10 | — |
| 284 | Grok-2xAI | 10.9 | 10.9 | 9.0 | — | — | — | — | — |
| 285 | Min istral 3 (3B Reasoning 2512)Mistral AI | 10.7 | 10.9 | 6.7 | — | — | — | — | — |
| 286 | DeepSeek R1 Distill Qwen 14BDeepSeek | 10.4 | 10.7 | 5.3 | — | — | — | — | — |
| 287 | Qwen3 VL 4B InstructAlibaba Cloud / Qwen Team | 10.2 | 7.7 | — | 2.1 | — | — | — | — |
| 288 | Magistral Small 2506Mistral AI | 10.2 | 10.4 | 5.0 | — | — | — | — | — |
| 289 | Mistral Large 3Mistral AI | 10.0 | 8.0 | — | — | — | — | — | — |
| 290 | Gemini 2.5 Flash-LiteGoogle | 9.9 | 10.4 | -3.1 | — | — | — | — | — |
| 291 | MiniStral 3 (14B Instruct 2512)Mistral AI | 9.8 | 11.6 | — | — | — | — | — | — |
| 292 | o1-miniOpenAI | 9.5 | 9.7 | 14.2 | — | — | — | — | — |
| 293 | Claude 3 OpusAnthropic | 9.4 | 9.3 | 5.0 | — | — | — | — | — |
| 294 | Qwen2.5 VL 32B InstructAlibaba Cloud / Qwen Team | 9.3 | 8.0 | 12.2 | -3.2 | — | — | — | — |
| 295 | Devstral Small 1.1Mistral AI | 8.6 | 8.6 | 0.2 | — | — | — | — | — |
| 296 | Nova 2 SonicAmazon | 8.6 | 11.1 | — | — | 1M | — | $0.33 / $2.75 | — |
| 297 | Mistral Large 3 (675B Instruct 2512)Mistral AI | 8.5 | 8.7 | 1.4 | — | 262K | 46 t/s | $0.5 / $1.5 | — |
| 298 | GPT-4 TurboOpenAI | 8.5 | 8.7 | 7.1 | — | 128K | 97 t/s | $10 / $30 | — |
| 299 | QwQ-32B-PreviewAlibaba Cloud / Qwen Team | 8.5 | 8.8 | 4.3 | — | — | — | — | — |
| 300 | Mistral Large 3 (675B Instruct 2512 NVFP4)Mistral AI | 8.4 | 8.6 | 1.2 | — | — | — | — | — |
| 301 | Mistral Large 3 (675B Base)Mistral AI | 8.3 | 8.5 | 1.1 | — | — | — | — | — |
| 302 | Mistral Large 3 (675B Instruct 2512 Eagle)Mistral AI | 8.2 | 8.4 | 1.0 | — | — | — | — | — |
| 303 | Grok-2 minixAI | 8.1 | 7.7 | 5.8 | — | — | — | — | — |
| 304 | Mistral Small 3.2 24B InstructMistral AI | 8.0 | 7.5 | — | — | 128K | — | $0.075 / $0.2 | — |
| 305 | DeepSeek R1 Distill Qwen 7BDeepSeek | 7.9 | 8.2 | 1.9 | — | — | — | — | — |
| 306 | Gemma 3 27BGoogle | 7.8 | 8.1 | 5.2 | — | 131K | — | $0.08 / $0.16 | — |
| 307 | DeepSeek-V2.5DeepSeek | 7.6 | 7.7 | 6.0 | — | — | — | — | — |
| 308 | QvQ-72B-PreviewAlibaba Cloud / Qwen Team | 7.4 | 7.8 | — | — | — | — | — | — |
| 309 | Llama 4 ScoutMeta | 7.3 | 6.1 | 0.2 | — | 327K | — | $0.1 / $0.3 | — |
| 310 | Ministral 3 (8B Instruct 2512)Mistral AI | 7.3 | 8.9 | — | — | — | — | — | — |
| 311 | Qwen2.5 32B InstructAlibaba Cloud / Qwen Team | 7.2 | 6.7 | 9.2 | — | — | — | — | — |
| 312 | Llama 3.1 70B InstructMeta | 7.1 | 5.9 | 3.5 | — | 131K | — | $0.4 / $0.4 | — |
| 313 | Nova LiteAmazon | 7.1 | 6.6 | 5.5 | 4.8 | — | — | — | — |
| 314 | Mistral Large 2Mistral AI | 6.9 | 6.9 | 12.9 | — | — | — | — | — |
| 315 | Phi 4 Mini ReasoningMicrosoft | 6.6 | 7.0 | — | — | — | — | — | — |
| 316 | DeepSeek R1 Distill Llama 8BDeepSeek | 6.5 | 6.8 | 2.4 | — | — | — | — | — |
| 317 | Gemma 4 E2BGoogle | 6.2 | 5.1 | — | -3.6 | — | — | — | — |
| 318 | Qwen2.5 VL 7B InstructAlibaba Cloud / Qwen Team | 6.1 | 2.6 | — | -3.2 | — | — | — | — |
| 319 | Gemini 1.5 FlashGoogle | 5.5 | 5.1 | -0.5 | — | — | — | — | — |
| 320 | Qwen2 72B InstructAlibaba Cloud / Qwen Team | 5.3 | 5.2 | 9.7 | — | — | — | — | — |
| 321 | Gemma 3 12BGoogle | 5.1 | 5.5 | 2.6 | — | 131K | — | $0.05 / $0.15 | — |
| 322 | Phi 4Microsoft | 4.9 | 6.1 | 3.3 | — | 16K | — | $0.07 / $0.14 | — |
| 323 | Llama 3.2 90B InstructMeta | 4.7 | 6.2 | — | — | — | — | — | — |
| 324 | Qwen2.5-Omni-7BAlibaba Cloud / Qwen Team | 4.7 | 0.9 | 1.0 | — | — | — | — | — |
| 325 | Claude 3.5 HaikuAnthropic | 4.6 | 4.8 | 3.9 | — | — | — | — | — |
| 326 | Llama 3.1 Nemotron Nano 8B V1NVIDIA | 4.3 | 4.8 | — | — | — | — | — | — |
| 327 | Hermes 3 405BNous Research | 3.9 | 3.3 | — | — | 131K | — | $1 / $1 | — |
| 328 | Mistral Small 3 24B InstructMistral AI | 3.7 | 3.3 | 4.5 | — | 32K | — | $0.05 / $0.08 | — |
| 329 | Mistral Small 3.1 24B InstructMistral AI | 3.7 | 3.8 | 9.6 | — | — | — | — | — |
| 330 | Hermes 3 70BNous Research | 3.6 | 0.8 | — | — | 131K | — | $0.7 / $0.7 | — |
| 331 | Qwen2.5 14B InstructAlibaba Cloud / Qwen Team | 3.3 | 2.8 | 3.8 | — | — | — | — | — |
| 332 | Qwen3.5-2BAlibaba Cloud / Qwen Team | 3.3 | 4.9 | — | -0.8 | — | — | — | — |
| 333 | Ministral 3 (14B Base 2512)Mistral AI | 3.2 | 3.0 | — | — | — | — | — | — |
| 334 | GPT-4o miniOpenAI | 3.1 | 3.6 | -0.8 | — | — | — | — | — |
| 335 | LFM2.5-VL-3BLiquid AI | 2.7 | -2.7 | — | -5.2 | — | — | — | — |
| 336 | GPT-4OpenAI | 2.5 | 2.5 | -5.3 | — | — | — | — | — |
| 337 | DeepSeek VL2DeepSeek | 2.5 | -2.3 | — | — | — | — | — | — |
| 338 | Phi-4-multimodal-instructMicrosoft | 2.3 | -0.6 | — | — | — | — | — | — |
| 339 | Qwen2.5 7B InstructAlibaba Cloud / Qwen Team | 2.0 | 2.1 | 3.0 | — | — | — | — | — |
| 340 | Nova MicroAmazon | 1.9 | 1.2 | 2.5 | — | — | — | — | — |
| 341 | Granite 3.3 8B InstructIBM | 1.9 | 2.4 | 11.0 | — | — | — | — | — |
| 342 | Ministral 3 (3B Instruct 2512)Mistral AI | 1.8 | 2.3 | — | — | — | — | — | — |
| 343 | Qwen2.5-Coder 32B InstructAlibaba Cloud / Qwen Team | 1.6 | 1.6 | 9.8 | — | — | — | — | — |
| 344 | Phi-3.5-MoE-instructMicrosoft | 1.4 | 1.0 | -3.9 | — | — | — | — | — |
| 345 | Claude 3 SonnetAnthropic | 1.2 | 1.2 | -1.5 | — | — | — | — | — |
| 346 | GPT-4.1 nanoOpenAI | 1.1 | 1.5 | -12.1 | — | 1M | 200 t/s | $0.1 / $0.4 | — |
| 347 | DeepSeek VL2 SmallDeepSeek | 0.8 | -4.4 | — | — | — | — | — | — |
| 348 | Jamba 1.5 LargeAI21 Labs | 0.5 | 0.5 | — | — | — | — | — | — |
| 349 | Gemini DiffusionGoogle | 0.4 | 0.4 | 2.5 | — | — | — | — | — |
| 350 | Llama 3.1 Nemotron 70B InstructNVIDIA | -0.1 | -0.4 | — | — | — | — | — | — |
| 351 | Mistral Small 3.1 24B BaseMistral AI | -0.1 | 0.1 | — | — | — | — | — | — |
| 352 | Codestral-22BMistral AI | -0.4 | -0.5 | 2.4 | — | — | — | — | — |
| 353 | Mistral Small 3 24B BaseMistral AI | -0.8 | -0.8 | — | — | — | — | — | — |
| 354 | Command R+Cohere | -0.9 | -1.2 | — | — | — | — | — | — |
| 355 | Granite 3.3 8B BaseIBM | -1.0 | -1.1 | 10.9 | — | — | — | — | — |
| 356 | Gemma 2 27BGoogle | -1.2 | -1.5 | -8.0 | — | — | — | — | — |
| 357 | Gemini 1.5 Flash 8BGoogle | -1.2 | -0.8 | — | — | — | — | — | — |
| 358 | Grok-1.5VxAI | -1.2 | -0.5 | — | — | — | — | — | — |
| 359 | Ministral 3 (8B Base 2512)Mistral AI | -1.5 | -1.6 | — | — | — | — | — | — |
| 360 | Llama 3.2 11B InstructMeta | -1.8 | -1.2 | — | — | — | — | — | — |
| 361 | Claude 3 HaikuAnthropic | -1.9 | -2.0 | 0.5 | — | — | — | — | — |
| 362 | Pixtral-12BMistral AI | -2.1 | -4.2 | -2.7 | — | — | — | — | — |
| 363 | Gemma 3 4BGoogle | -2.3 | -3.4 | -4.5 | — | 131K | — | $0.05 / $0.1 | — |
| 364 | Shieldstral 1.0 (3B)Mistral AI | -2.7 | — | — | — | — | — | — | — |
| 365 | Llama 3.1 8B InstructMeta | -3.1 | -4.1 | -2.4 | — | 131K | — | $0.02 / $0.04 | — |
| 366 | GPT-5.1 Codex MiniOpenAI | -3.1 | -3.0 | — | — | — | — | — | — |
| 367 | Grok-1.5xAI | -3.1 | -2.7 | -1.0 | — | — | — | — | — |
| 368 | DeepSeek R1 Distill Qwen 1.5BDeepSeek | -3.5 | -3.1 | -4.7 | — | — | — | — | — |
| 369 | Qwen2.5-Coder 7B InstructAlibaba Cloud / Qwen Team | -3.6 | -3.6 | 6.3 | — | — | — | — | — |
| 370 | Grok-4.20 Beta Non-ReasoningxAI | -3.7 | — | — | — | — | — | — | — |
| 371 | Phi-3.5-vision-instructMicrosoft | -3.8 | -5.2 | — | — | — | — | — | — |
| 372 | Phi-3.5-mini-instructMicrosoft | -4.3 | -5.2 | -6.9 | — | — | — | — | — |
| 373 | Phi 4 MiniMicrosoft | -4.5 | -4.6 | — | — | — | — | — | — |
| 374 | MedGemma 4B ITGoogle | -4.8 | -5.2 | — | — | — | — | — | — |
| 375 | Qwen2 7B InstructAlibaba Cloud / Qwen Team | -5.0 | -4.7 | 0.8 | — | — | — | — | — |
| 376 | Gemma 3n E4B InstructedGoogle | -5.1 | -6.0 | -0.3 | — | — | — | — | — |
| 377 | DeepSeek VL2 TinyDeepSeek | -5.1 | -13.1 | — | — | — | — | — | — |
| 378 | Ministral 8B InstructMistral AI | -5.1 | -5.2 | -14.5 | — | — | — | — | — |
| 379 | Mistral NeMo InstructMistral AI | -5.2 | -5.4 | — | — | 131K | — | $0.019 / $0.03 | — |
| 380 | Gemma 2 9BGoogle | -5.3 | -5.6 | -11.1 | — | — | — | — | — |
| 381 | North Micro Vision InstructCohere | -5.6 | -8.2 | — | — | — | — | — | — |
| 382 | Gemma 3n E4B Instructed LiteRT PreviewGoogle | -5.6 | -6.4 | -0.2 | — | — | — | — | — |
| 383 | Gemini 1.0 ProGoogle | -5.9 | -8.5 | — | — | — | — | — | — |
| 384 | IBM Granite 4.0 Tiny PreviewIBM | -5.9 | -5.9 | 2.8 | — | — | — | — | — |
| 385 | Jamba 1.5 MiniAI21 Labs | -6.1 | -6.1 | — | — | — | — | — | — |
| 386 | Llama 3.2 3B InstructMeta | -6.5 | -7.2 | — | — | — | — | — | — |
| 387 | Gemma 3n E4BGoogle | -6.6 | -6.9 | — | — | — | — | — | — |
| 388 | Qwen3.5-0.8BAlibaba Cloud / Qwen Team | -6.7 | -4.0 | — | -9.7 | — | — | — | — |
| 389 | Ministral 3 (3B Base 2512)Mistral AI | -6.8 | -6.8 | — | — | — | — | — | — |
| 390 | GPT-3.5 TurboOpenAI | -9.9 | -9.2 | -4.6 | — | 16K | 100 t/s | $0.5 / $1.5 | — |
| 391 | Gemma 3n E2B Instructed LiteRT (Preview)Google | -10.1 | -10.3 | -4.5 | — | — | — | — | — |
| 392 | Gemma 3n E2BGoogle | -10.7 | -10.9 | — | — | — | — | — | — |
| 393 | Gemma 3 1BGoogle | -11.0 | -12.6 | -10.3 | — | — | — | — | — |
| 394 | Gemma 3n E2B InstructedGoogle | -11.1 | -11.4 | -4.6 | — | — | — | — | — |
| 395 | ERNIE 4.5Baidu | -13.9 | -13.8 | — | — | — | — | — | — |
В IRI Gateway уже есть Claude Opus 5.5, Claude Sonnet 5.5, GPT-6 Astra, Gemini 4 Argon, GPT-6.1 Sol, Claude Fable 5.1. Один ключ на все модели, OpenAI-совместимый API, оплата в рублях, в том числе по счёту для юрлица. IRI Zero — бесплатный режим на доступной бесплатной вычислительной ёмкости. Если бесплатная ёмкость временно исчерпана, запрос не переключается на платную модель автоматически.
Разборы, в которых сравниваем модели на конкретных задачах: код, агенты, длинный контекст и цена.
По сводному баллу сейчас лидирует Claude Opus 5.5, за ней Claude Sonnet 5.5, GPT-6 Astra. Но лучшая модель зависит от задачи: для кода, агентов и массовых задач лидеры разные — смотрите блок «Лидеры по задачам».
По баллу coding впереди Claude Opus 5.5. Для повседневного кода обычно хватает моделей из первой двадцатки подешевле — сравните цену за задачу.
Через IRI Gateway: один API-ключ, оплата в рублях, в том числе по счёту для юрлица. Из топ-10 рейтинга в Gateway доступны 6 моделей.
Да, многие модели с открытыми весами можно скачать или попробовать бесплатно, а в IRI Gateway доступен IRI Zero — бесплатный режим на доступной бесплатной вычислительной ёмкости; если ёмкость исчерпана, запрос не переключается на платную модель автоматически. Подборка — в статье «Лучшие нейросети 2026».
Сводный балл собран по открытым бенчмаркам LLM Stats: рассуждения, код, работа агентом. Обновляем рейтинг каждую неделю, дата — вверху страницы.