Skip to content

Benchmark

GPQA Diamond

Science benchmarks measure how well models answer science questions. Scores are percentages. Higher is better.

Models and systems measured
282
Scores recorded
1623
Sources
4
IQ weight
1.72

Score distribution

79.3%
Mean
84.5%
Median
15.7%
Std dev
56.8%
10th percentile
92.2%
90th percentile
0.168%
Min
0.168%96.8%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.72 for chat models, measured over 272 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#Model or system
951MiMo V2.5OpenRouter harness, xiaomi/mimo-v2.5-20260422, Novita, public 90-day aggregate· OpenRouter benchmark exports82.8%
952GLM 4.7OpenRouter harness, z-ai/glm-4.7-20251222, StreamLake, public 90-day aggregate· OpenRouter benchmark exports82.8%
953GPT-5.6 Solnone· Epoch AI Benchmarking Hub data82.8%
954GLM 5.3 FlashOpenRouter harness, z-ai/glm-5.3-flash-20260826, Decart, public 90-day aggregate· OpenRouter benchmark exports82.8%
955GPT 5.4 Nanoxhigh · published setting· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.82.8%
956DeepSeek V3.2OpenRouter harness, deepseek/deepseek-v3.2-20251201, Google, public 90-day aggregate· OpenRouter benchmark exports82.8%
957GLM 4.7OpenRouter harness, z-ai/glm-4.7-20251222, DeepInfra, public 90-day aggregate· OpenRouter benchmark exports82.8%
958Qwen3.6 35B A3BOpenRouter harness, qwen/qwen3.6-35b-a3b-20260415, Phala, public 90-day aggregate· OpenRouter benchmark exports82.8%
959Qwen 3.8 27BOpenRouter harness, qwen/qwen3.8-27b-20260814, Ionstream, public 90-day aggregate· OpenRouter benchmark exports82.8%
960DeepSeek V3.2OpenRouter harness, deepseek/deepseek-v3.2-20251201, DigitalOcean, public 90-day aggregate· OpenRouter benchmark exports82.7%
961Qwen 3.8 27BOpenRouter harness, qwen/qwen3.8-27b-20260814, AkashML, public 90-day aggregate· OpenRouter benchmark exports82.7%
962Qwen 3.8 27BOpenRouter harness, qwen/qwen3.8-27b-20260814, Novita, public 90-day aggregate· OpenRouter benchmark exports82.7%
963GPT 5.2low· Epoch AI Benchmarking Hub data82.7%
964Kimi K2.7 CodeOpenRouter harness, moonshotai/kimi-k2.7-code-20260612, BaseTen, public 90-day aggregate· OpenRouter benchmark exports82.7%
965Qwen 3.8 27BOpenRouter harness, qwen/qwen3.8-27b-20260814, CoreWeave, public 90-day aggregate· OpenRouter benchmark exports82.7%
966DeepSeek V3.2OpenRouter harness, deepseek/deepseek-v3.2-20251201, Alibaba, public 90-day aggregate· OpenRouter benchmark exports82.7%
967DeepSeek V3.1 TerminusOpenRouter harness, deepseek/deepseek-v3.1-terminus, Novita, public 90-day aggregate· OpenRouter benchmark exports82.7%
968Muse Glimmer 30BOpenRouter harness, meta/muse-glimmer-30b-20260810, Parasail, public 90-day aggregate· OpenRouter benchmark exports82.7%
969GLM 5.3OpenRouter harness, z-ai/glm-5.3-20260816, Friendli, public 90-day aggregate· OpenRouter benchmark exports82.6%
970GLM 5.3 FlashOpenRouter harness, z-ai/glm-5.3-flash-20260826, CoreWeave, public 90-day aggregate· OpenRouter benchmark exports82.6%
971GLM 5.3OpenRouter harness, z-ai/glm-5.3-20260816, Io Net, public 90-day aggregate· OpenRouter benchmark exports82.6%
972Qwen 3.8 27BOpenRouter harness, qwen/qwen3.8-27b-20260814, Chutes, public 90-day aggregate· OpenRouter benchmark exports82.6%
973GLM 5.3 FlashOpenRouter harness, z-ai/glm-5.3-flash-20260826, NextBit, public 90-day aggregate· OpenRouter benchmark exports82.6%
974GLM 5.3 FlashOpenRouter harness, z-ai/glm-5.3-flash-20260826, Friendli, public 90-day aggregate· OpenRouter benchmark exports82.6%
975DeepSeek V4 Pro 0813OpenRouter harness, deepseek/deepseek-v4-pro-20260813, Venice, public 90-day aggregate· OpenRouter benchmark exports82.6%
976Gemini 2.5 ProOpenRouter harness, google/gemini-2.5-pro, auto-routing, public 90-day aggregate· OpenRouter benchmark exports82.6%
977Gemma 4 31B ITOpenRouter harness, google/gemma-4-31b-it-20260402, auto-routing, public 90-day aggregate· OpenRouter benchmark exports82.5%
978GLM 5.3 FlashOpenRouter harness, z-ai/glm-5.3-flash-20260826, Modal, public 90-day aggregate· OpenRouter benchmark exports82.5%
979Kimi K2.6OpenRouter harness, moonshotai/kimi-k2.6-20260420, AtlasCloud, public 90-day aggregate· OpenRouter benchmark exports82.5%
980Qwen3.5 122B-A10BOpenRouter harness, qwen/qwen3.5-122b-a10b-20260224, SiliconFlow, public 90-day aggregate· OpenRouter benchmark exports82.5%
981GPT-5.5 Instant· Epoch AI Benchmarking Hub data82.5%
982MiMo V2.5OpenRouter harness, xiaomi/mimo-v2.5-20260422, Xiaomi, public 90-day aggregate· OpenRouter benchmark exports82.5%
983GLM 5.1OpenRouter harness, z-ai/glm-5.1-20260406, SiliconFlow, public 90-day aggregate· OpenRouter benchmark exports82.4%
984Gemini 2.5 ProOpenRouter harness, google/gemini-2.5-pro, Google, public 90-day aggregate· OpenRouter benchmark exports82.4%
985DeepSeek V3.2· OpenEvals Leaderboard data· Reported by the model's developer. The source relaying it did not verify it.82.4%
986DeepSeek V3.2Thinking mode · temperature 1.0 · 128K context · pass@1· Reported by DeepSeek· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.82.4%
987DeepSeek V3.2OpenRouter harness, deepseek/deepseek-v3.2-20251201, auto-routing, public 90-day aggregate· OpenRouter benchmark exports82.4%
988Qwen3.5 35B A3BOpenRouter harness, qwen/qwen3.5-35b-a3b-20260224, DeepInfra, public 90-day aggregate· OpenRouter benchmark exports82.3%
989Claude Sonnet 4.559k· Epoch AI Benchmarking Hub data82.3%
990GPT-5.6 Lunalow· Epoch AI Benchmarking Hub data82.3%
991Qwen3.5 Flash· Epoch AI Benchmarking Hub data82.3%
992Qwen3.7 Flash· Epoch AI Benchmarking Hub data82.3%
993Qwen 3.8 27BOpenRouter harness, qwen/qwen3.8-27b-20260814, Nebius, public 90-day aggregate· OpenRouter benchmark exports82.3%
994Gemma 4 26B A4BGemma 4 Technical Report · Table 5 · thinking mode unless noted· Reported by Google DeepMind· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.82.3%
995GLM 5.3OpenRouter harness, z-ai/glm-5.3-20260816, Featherless, public 90-day aggregate· OpenRouter benchmark exports82.3%
996Gemini 2.5 ProOpenRouter harness, google/gemini-2.5-pro, Google AI Studio, public 90-day aggregate· OpenRouter benchmark exports82.2%
997DeepSeek V3.2OpenRouter harness, deepseek/deepseek-v3.2-20251201, Baidu, public 90-day aggregate· OpenRouter benchmark exports82.2%
998Qwen 3.8 27BOpenRouter harness, qwen/qwen3.8-27b-20260814, auto-routing, public 90-day aggregate· OpenRouter benchmark exports82.2%
999GLM 5.3 FlashOpenRouter harness, z-ai/glm-5.3-flash-20260826, Cohere, public 90-day aggregate· OpenRouter benchmark exports82.2%
1000Qwen 3.8 27BOpenRouter harness, qwen/qwen3.8-27b-20260814, GMICloud, public 90-day aggregate· OpenRouter benchmark exports82.2%

Showing 951–1,000 of 1,621 results.

Not ranked

The source marks these results as descriptive, not capability scores, so we list them by name instead of ranking them.

Model or system
MiMo V2 Flash Baseofficial Base table · 5-shot· Reported by Xiaomi MiMo· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.55.1%
Qwen3.8-2.4T-A95Blifecycle-page core-benchmark claim· Reported by Alibaba Cloud Model Studio· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.92.6%

Tokens the model generated per task, as measured by the source. Comparable within this benchmark only.

Data credits: Epoch AI Benchmarking Hub, CC BY 4.0 (licence) · BenchBoard by SYuan03 (github.com/SYuan03/benchboard). MIT covers its code and data organization; no results licence is inferred. Numeric relays retain the original stating-source links. · OpenRouter public benchmark leaderboard and search-lane exports, CC BY 4.0. Fractional source scores converted to percentages; immutable dated snapshots. (licence) · OpenEvals/leaderboard-data, MIT (licence)