LiveBench Leaderboard
reasoning/coding76 models · Updated October 2026
On the LiveBench benchmark, GPT-5.5 ranks #1 with a score of 81.3, while DeepSeek V4 Flash 0423 offers the best score-per-dollar at $0.18/1M output tokens. The full ranking, with cost per million tokens, is below.
💰 Best value
DeepSeek V4 Flash 0423 — score 65.5 at $0.18/1M output tokens
| # | Model | Score | Output / 1M | Context |
|---|---|---|---|---|
| 1 | GPT-5.5OpenAI | 81.3 | $30.00 | 1.1M |
| 2 | GPT-5.4OpenAI | 80.9 | $15.00 | 1.1M |
| 3 | Gemini-3.1-ProPoe | 80.7 | $12.12 | 1M |
| 4 | Claude Opus 4.8 ThinkingNanoGPT | 79.5 | $25.00 | 1M |
| 5 | Claude Fable 5Anthropic | 78.6 | $50.00 | 1M |
| 6 | Gemini 3.1 Pro PreviewGoogle | 77.1 | $12.00 | 1M |
| 7 | Claude Opus 4.7Anthropic | 77.1 | $25.00 | 1M |
| 8 | Claude Opus 4.6 (Fast)Anthropic | 76.8 | $150.00 | 1M |
| 9 | Claude Opus 4.5Anthropic | 76.0 | $25.00 | 200K |
| 10 | Gemini 3.5 FlashGoogle | 75.8 | $9.00 | 1M |
| 11 | Claude Sonnet 4.6 ThinkingNanoGPT | 75.7 | $15.00 | 1M |
| 12 | GPT-5.2OpenAI | 75.4 | $14.00 | 400K |
| 13 | Qwen3.7 Max ThinkingNanoGPT | 75.2 | $7.50 | 1M |
| 14 | DeepSeek V4 ProCrofAI | 74.4 | $0.80 | 1M |
| 15 | GPT-5.1-Codex-MaxOpenAI | 74.4 | $10.00 | 400K |
| 16 | GPT-5.2-CodexOpenAI | 74.3 | $14.00 | 400K |
| 17 | Gemini 3 ProOpenCode | 73.5 | $12.00 | 1M |
| 18 | GPT-5.3-CodexOpenAI | 73.2 | $14.00 | 400K |
| 19 | GLM 5.2Z.ai | 73.2 | $7.00 | 1M |
| 20 | Qwen3.7 MaxQwen | 73.1 | $4.43 | 1M |
| 21 | Gemini 3 Flash ThinkingNanoGPT | 73.0 | $3.00 | 1M |
| 22 | GPT-5.1OpenAI | 72.6 | $10.00 | 400K |
| 23 | Kimi K2.6MoonshotAI | 72.4 | $3.41 | 262K |
| 24 | DeepSeek V4 Pro 0423DeepSeek | 71.6 | $1.91 | 1M |
| 25 | GPT-5.4 NanoOpenAI | 71.3 | $1.25 | 400K |
| 26 | GPT-5 ProOpenAI | 71.3 | $120.00 | 400K |
| 27 | Qwen3.6 PlusQwen | 70.8 | $1.95 | 1M |
| 28 | GLM 5.1Z.ai | 70.6 | $3.04 | 205K |
| 29 | MiniMax M3 ThinkingNanoGPT | 70.0 | $1.20 | 512K |
| 30 | Grok Build 0.1SpaceXAI | 69.6 | $2.00 | 256K |
| 31 | GPT-5.1-CodexOpenAI | 69.3 | $10.00 | 400K |
| 32 | Kimi K2.5MoonshotAI | 69.2 | $3.00 | 262K |
| 33 | Grok 4.20 (Reasoning)xAI | 69.0 | $2.50 | 1M |
| 34 | GLM 5Z.ai | 68.7 | $3.20 | 205K |
| 35 | Kimi K2.7 CodeMoonshotAI | 68.4 | $3.35 | 262K |
| 36 | Claude Sonnet 4.5Anthropic | 67.9 | $15.00 | 1M |
| 37 | GPT-5.4 MiniOpenAI | 67.7 | $4.50 | 400K |
| 38 | DeepSeek V4 Flash LatestDeepSeek | 67.7 | $0.58 | 1M |
| 39 | Grok 4.3SpaceXAI | 67.4 | $2.50 | 1M |
| 40 | MiniMax M3MiniMax | 67.3 | $1.20 | 1M |
| 41 | GPT-5 MiniOpenAI | 66.6 | $2.00 | 400K |
| 42 | DeepSeek V4 Flash 0423DeepSeek | 65.5 | $0.18 | 1M |
| 43 | MiniMax M2.7MiniMax | 65.0 | $1.20 | 205K |
| 44 | Qwen3.6 27BQwen | 64.0 | $3.20 | 262K |
| 45 | DeepSeek V3.2DeepSeek | 63.1 | $0.80 | 164K |
| 46 | Gemma 4 31B ITQVAC | 62.4 | $0.00 | 262K |
| 47 | Kimi K2 ThinkingMoonshotAI | 62.3 | $2.50 | 262K |
| 48 | Gemini 3.1 Flash LiteGoogle | 62.1 | $1.50 | 1M |
| 49 | grok-4-0709Jiekou.AI | 61.8 | $13.50 | 256K |
| 50 | Claude 4.1 Opus Thinking (32K)NanoGPT | 61.4 | $75.00 | 200K |
| 51 | Claude Haiku 4.5Anthropic | 61.0 | $5.00 | 200K |
| 52 | GPT-5.1-Codex-MiniOpenAI | 60.8 | $2.00 | 400K |
| 53 | Claude 4 SonnetNanoGPT | 60.6 | $14.99 | 200K |
| 54 | Qwen3.6 FlashQwen | 60.5 | $1.13 | 1M |
| 55 | MiniMax M2.5MiniMax | 60.3 | $0.95 | 205K |
| 56 | Grok 4.1 FastOfox | 60.1 | $0.50 | 2M |
| 57 | GPT-5.3-InstantPoe | 59.8 | $13.00 | 128K |
| 58 | MiMo-V2-ProXiaomi | 58.4 | $0.87 | 1M |
| 59 | Gemini 2.5 Pro Preview 06-05Google | 57.5 | $10.00 | 1M |
| 60 | GLM 4.7Z.ai | 57.3 | $2.20 | 205K |
| 61 | GLM 4.6Z.ai | 54.7 | $2.00 | 205K |
| 62 | Qwen3 235B A22B Thinking 2507Qwen | 52.9 | $3.50 | 128K |
| 63 | Gemini 2.5 Flash PreviewNanoGPT | 52.3 | $0.60 | 1M |
| 64 | Qwen3 Next 80B A3B ThinkingQwen | 51.0 | $1.20 | 262K |
| 65 | Nemotron 3 UltraNVIDIA | 50.7 | $2.40 | 262K |
| 66 | GPT-5 NanoOpenAI | 48.8 | $0.40 | 400K |
| 67 | GLM 5V Turbo ThinkingNanoGPT | 48.8 | $4.00 | 203K |
| 68 | GPT OSS 120BNeon | 46.4 | $0.60 | 131K |
| 69 | Qwen3 32BQwen | 42.7 | $0.28 | 131K |
| 70 | Gemini 2.5 Flash LiteGoogle | 41.5 | $0.40 | 1M |
| 71 | GLM 4.6VZ.ai | 38.9 | $0.90 | 131K |
| 72 | Qwen3 30B A3BQwen | 38.8 | $0.50 | 41K |
| 73 | Devstral 2 2512Mistral | 38.8 | $2.00 | 262K |
| 74 | Grok 4.20 (Non-Reasoning)xAI | 37.9 | $2.50 | 1M |
| 75 | Nemotron 3 Super 120B A12BSynthetic | 32.0 | $1.00 | 262K |
| 76 | xAI Grok 4.1 Fast Non-ReasoningHelicone | 31.6 | $0.50 | 2M |
Frequently asked questions
What is the LiveBench benchmark?
LiveBench is a reasoning/coding used to evaluate and rank large language models. This page lists 76 models by their LiveBench score.
Which model is best on LiveBench?
GPT-5.5 (OpenAI) currently leads the LiveBench leaderboard with a score of 81.3.
What is the best value model on LiveBench?
DeepSeek V4 Flash 0423 offers the best score-per-dollar on LiveBench — a score of 65.5 at $0.18 per 1M output tokens.
Pricing is indicative — confirm with the provider before production use. Updated October 2026.