Skip to content

Benchmark

Dubesor STEM

Scores are percentages.

Models measured
225
Scores recorded
268
Sources
1
IQ weight
1.64

Score distribution

51.4%
Mean
55.8%
Median
23.1%
Std dev
15.8%
10th percentile
79.9%
90th percentile
0.5%
Min
0.5%93.8%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.64 for chat models, measured over 225 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
51GLM 5thinking· Dubesor LLM Benchmark73.3%
52Kimi K2.5thinking· Dubesor LLM Benchmark73.3%
53ChatGPT-4o Latest· Dubesor LLM Benchmark73.1%
54Gemini 3 Flashthinking· Dubesor LLM Benchmark73.1%
55Gemini 2.5 Pro· Dubesor LLM Benchmark72.6%
56Gemini 3.1 Flash Lite· Dubesor LLM Benchmark72.6%
57Grok 4 Fastreasoning· Dubesor LLM Benchmark72.5%
58Gemma 4 26B A4Bthinking· Dubesor LLM Benchmark71.9%
59DeepSeek V3.1· Dubesor LLM Benchmark71.5%
60gemini-1.5-pro-002· Dubesor LLM Benchmark71.2%
61DeepSeek V3.1thinking· Dubesor LLM Benchmark71.1%
62Qwen3 32Bthinking· Dubesor LLM Benchmark70.8%
63Qwen3.5 Plusthinking· Dubesor LLM Benchmark70.7%
64Kimi K2 Thinking· Dubesor LLM Benchmark70.4%
65Qwen3 30B A3Bthinking· Dubesor LLM Benchmark70.4%
66DeepSeek V3· Dubesor LLM Benchmark70.3%
67Llama 3.1 Nemotron Ultra 253B v1thinking· Dubesor LLM Benchmark70.2%
68Ling 1t· Dubesor LLM Benchmark70%
69Qwen3.5 397B A17B· Dubesor LLM Benchmark69.7%
70Gemini 3 Flash· Dubesor LLM Benchmark69.6%
71GPT-5.1 Codex· Dubesor LLM Benchmark68.5%
72Qwen3.5 397B A17Bthinking· Dubesor LLM Benchmark68.1%
73Kimi K2 Instruct· Dubesor LLM Benchmark68%
74Llama 3.3 70B Instruct· Dubesor LLM Benchmark67.9%
75Nemotron 3 Super 120B A12Bthinking· Dubesor LLM Benchmark67.7%
76Phi 4 Reasoning Plus· Dubesor LLM Benchmark67.7%
77Qwen 3 235B A22B Thinking 2507· Dubesor LLM Benchmark67.6%
78Qwen3.5 Plus· Dubesor LLM Benchmark67.5%
79Grok 3 Mini· Dubesor LLM Benchmark67.3%
80Llama 3.1 Nemotron Ultra 253B v1· Dubesor LLM Benchmark66.6%
81Gemini 2.5 Flashthinking· Dubesor LLM Benchmark66.1%
82Llama 3.3 Nemotron Super 49B v1thinking· Dubesor LLM Benchmark66.1%
83GPT-5 nano· Dubesor LLM Benchmark65.7%
84Gemma 4 26B A4B· Dubesor LLM Benchmark65.2%
85MiniMax M2.1· Dubesor LLM Benchmark64.9%
86Qwen3 Next 80B A3Bthinking· Dubesor LLM Benchmark64.7%
87MiniMax M2· Dubesor LLM Benchmark64.6%
88Mistral Large 3· Dubesor LLM Benchmark64.6%
89Qwen3 Coder Plus· Dubesor LLM Benchmark64.6%
90GPT OSS 20b· Dubesor LLM Benchmark64.2%
91Llama 3.3 Nemotron Super 49B v1.5· Dubesor LLM Benchmark64.2%
92gemini-2.5-flash-preview-04-17· Dubesor LLM Benchmark64%
93GPT OSS 120B· Dubesor LLM Benchmark64%
94Claude Haiku 4.5thinking· Dubesor LLM Benchmark63.8%
95Qwen3.5 27Bthinking· Dubesor LLM Benchmark63.8%
96GPT-5 Codex· Dubesor LLM Benchmark63%
97Llama 3.1 Nemotron 70B Instruct· Dubesor LLM Benchmark61.9%
98Qwen3-4B-Thinking-2507· Dubesor LLM Benchmark61.8%
99Qwen3 30B A3B· Dubesor LLM Benchmark61.6%
100LongCat Flash Chat· Dubesor LLM Benchmark61.5%

Showing 51–100 of 268 results.