llmrun
Math

LiveBench Math Leaderboard

LiveBench Math measures mathematical problem-solving on contamination-free, regularly-refreshed questions, including competition-style problems.

Source: livebench16 open models ranked+38 proprietaryData through Jun 2026

Open modelsAll models

All models ranked on LiveBench Math

Proprietary / closed models are shown dimmed — you can't run them locally, but they show where the open field stands.

#ModelScore
1Claude Fable 5.1 Max · proprietary
97.0
2GPT 6 Astra Max · proprietary
96.8
3GPT 5.6 Sol Max · proprietary
96.2
4Claude Fable 5 Max · proprietary
96.0
5Muse Spark 1.3 · proprietary
96.0
6GPT 5.5 · proprietary
95.9
7Claude Opus 5 Max · proprietary
95.7
8DeepSeek V4 Pro 0813 · 1650.5B
95.1
9GPT 5.6 Terra Max · proprietary
94.9
10Claude Opus 4.8 Max · proprietary
94.3
11GPT 5.4 · proprietary
94.2
12Gemini 3.7 Flash · proprietary
93.5
13GPT 5.2 · proprietary
93.2
14Claude Sonnet 5 · proprietary
92.9
15Claude Opus 4.7 · proprietary
92.8
16Grok 4.6 · proprietary
92.6
17Gemini 3.8 Flash · proprietary
91.6
18Qwen3.8 Max · proprietary
91.3
19Muse Spark 1.2 · proprietary
91.2
20Gemini 3.1 Pro · proprietary
91.0
21GPT 5.4 Nano · proprietary
91.0
22Grok 4.5 · proprietary
90.8
23DeepSeek V4 Pro · 1598.8B
90.7
24Claude Opus 4.5 · proprietary
90.4
25GLM 5.2 · 753.3B
89.8
26Claude Opus 4.6 · proprietary
89.3
27GPT 5.2 Codex · proprietary
88.8
28NVIDIA Nemotron 3 Ultra 550B A55B BF16 · 560.5B
88.7
29Inkling · 952.4B
88.4
30Gemini 3.5 Flash · proprietary
88.2
31GLM 5.3 · 753.3B
87.9
32DeepSeek v4 Flash Vision · proprietary
87.8
33GPT 5.6 Luna Max · proprietary
87.2
34Muse Spark 1.1 · proprietary
87.1
35Claude Sonnet 4.6 · proprietary
87.0
36DeepSeek V4 Flash 0731 · 304.2B
86.8
37Gemini 3.6 Flash · proprietary
86.4
38Qwen3.8 27B · 27.8B
86.2
39Qwen3.8 Flash Next · 180.0B
85.8
40Qwen3.7 Max · proprietary
85.3
41Kimi K3 · 2779.9B
84.4
42Grok 4.3 · proprietary
84.3
43Kimi K2.6 · 1026.9B
84.3
44Smaug Agentic · proprietary
83.9
45Qwen3.6 Plus · proprietary
83.7
46GLM 5.3 Flash · 321.3B
81.2
47Qwen3.6 27B · 27.8B
79.9
48DeepSeek V4 Flash · 290.9B
79.7
49Kimi K2.7 Code · 1026.9B
79.6
50GPT 5.4 Mini · proprietary
78.5
51Grok Build 0.1 · proprietary
78.4
52Ox Alpha Max · proprietary
77.5
53MiniMax M3 · 427.0B
77.0
54Gemini 3.5 Flash Lite · proprietary
73.7

Score vs model size

Which models give the most quality for their size — the ones worth running locally.

100B1Tmodel size (log scale) →95.177.0Inkling · 952B · 88.4GLM 5.3 · 753B · 87.9Qwen3.8 Flash Next · 180B · 85.8Kimi K3 · 2.8T · 84.4Kimi K2.6 · 1T · 84.3GLM 5.3 Flash · 321B · 81.2Qwen3.6 27B · 28B · 79.9DeepSeek V4 Flash · 291B · 79.7Kimi K2.7 Code · 1T · 79.6MiniMax M3 · 427B · 77.0Qwen3.8 27B · 28B · 86.2Qwen3.8 27BDeepSeek V4 Flash 0731 · 304B · 86.8DeepSeek V4 Flash 0731NVIDIA Nemotron 3 Ultra 550B A55B BF16 · 561B · 88.7NVIDIA Nemotron 3 Ult…GLM 5.2 · 753B · 89.8GLM 5.2DeepSeek V4 Pro · 1.6T · 90.7DeepSeek V4 ProDeepSeek V4 Pro 0813 · 1.7T · 95.1DeepSeek V4 Pro 0813
Each dot is a model. Up = higher score, left = smaller (easier to run locally). The dashed line marks the efficiency frontier — the best score you can get at each size or smaller.
  • Qwen3.8 27B, 28B, score 86.2 — on the efficiency frontier (best score at its size or smaller).
  • DeepSeek V4 Flash 0731, 304B, score 86.8 — on the efficiency frontier (best score at its size or smaller).
  • NVIDIA Nemotron 3 Ultra 550B A55B BF16, 561B, score 88.7 — on the efficiency frontier (best score at its size or smaller).
  • GLM 5.2, 753B, score 89.8 — on the efficiency frontier (best score at its size or smaller).
  • DeepSeek V4 Pro, 1.6T, score 90.7 — on the efficiency frontier (best score at its size or smaller).
  • DeepSeek V4 Pro 0813, 1.7T, score 95.1 — on the efficiency frontier (best score at its size or smaller).

LiveBench Math: frequently asked questions

What is the best open LLM on LiveBench Math?
DeepSeek V4 Pro 0813 is the top open model on LiveBench Math, scoring 95.1. Among all models tested — including proprietary ones — it ranks #8. The top model overall is Claude Fable 5.1 Max (Anthropic) at 97.0.
What's the best LiveBench Math model you can run on a 24 GB GPU?
Qwen3.8 27B is the highest-scoring open model that fits in 24 GB at 4-bit quantization (about 15 GB), scoring 86.2 on LiveBench Math.
Can open models match proprietary models on LiveBench Math?
Not quite on LiveBench Math: the strongest proprietary model (Claude Fable 5.1 Max) scores 97.0, ahead of the best open model (DeepSeek V4 Pro 0813) at 95.1 — but you can run the open one yourself.

Scores aggregated from livebench. llmrun does not run this benchmark — see the source for methodology, or the about benchmarks for what it measures.