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
801GLM 5.3 FlashOpenRouter harness, z-ai/glm-5.3-flash-20260826, Novita, public 90-day aggregate· OpenRouter benchmark exports84.7%
802Qwen3.5 122B-A10BOpenRouter harness, qwen/qwen3.5-122b-a10b-20260224, Novita, public 90-day aggregate· OpenRouter benchmark exports84.7%
803Qwen3.6 27BOpenRouter harness, qwen/qwen3.6-27b-20260422, Phala, public 90-day aggregate· OpenRouter benchmark exports84.7%
804Gemma 4 31B ITOpenRouter harness, google/gemma-4-31b-it-20260402, Chutes, public 90-day aggregate· OpenRouter benchmark exports84.7%
805GLM 5.2OpenRouter harness, z-ai/glm-5.2-20260616, Nebius, public 90-day aggregate· OpenRouter benchmark exports84.7%
806Kimi K2.6OpenRouter harness, moonshotai/kimi-k2.6-20260420, StreamLake, public 90-day aggregate· OpenRouter benchmark exports84.7%
807GLM 5.3OpenRouter harness, z-ai/glm-5.3-20260816, Phala, public 90-day aggregate· OpenRouter benchmark exports84.7%
808Gemma 4 31B ITOpenRouter harness, google/gemma-4-31b-it-20260402, KT, public 90-day aggregate· OpenRouter benchmark exports84.6%
809MiniMax M2.5OpenRouter harness, minimax/minimax-m2.5-20260211, Minimax, public 90-day aggregate· OpenRouter benchmark exports84.6%
810Nemotron 3.5 Super VLOpenRouter harness, nvidia/nemotron-3.5-super-vl-20261015, Crusoe, public 90-day aggregate· OpenRouter benchmark exports84.6%
811GLM 5.3OpenRouter harness, z-ai/glm-5.3-20260816, Fireworks, public 90-day aggregate· OpenRouter benchmark exports84.6%
812GLM 5.3OpenRouter harness, z-ai/glm-5.3-20260816, Crusoe, public 90-day aggregate· OpenRouter benchmark exports84.5%
813GLM 5.3 FlashOpenRouter harness, z-ai/glm-5.3-flash-20260826, DekaLLM, public 90-day aggregate· OpenRouter benchmark exports84.5%
814GLM 5.3 FlashOpenRouter harness, z-ai/glm-5.3-flash-20260826, Inceptron, public 90-day aggregate· OpenRouter benchmark exports84.5%
815MiniMax M2.5OpenRouter harness, minimax/minimax-m2.5-20260211, SiliconFlow, public 90-day aggregate· OpenRouter benchmark exports84.5%
816Gemma 4 31B ITOpenRouter harness, google/gemma-4-31b-it-20260402, Novita, public 90-day aggregate· OpenRouter benchmark exports84.5%
817Kimi K2 Thinking· OpenEvals Leaderboard data· Reported by the model's developer. The source relaying it did not verify it.
Values from 2 sources
84.5%
818Inkling FP4OpenRouter harness, thinkingmachines/inkling-20260715, DeepInfra, public 90-day aggregate· OpenRouter benchmark exports84.5%
819GLM 4.7OpenRouter harness, z-ai/glm-4.7-20251222, Z.AI, public 90-day aggregate· OpenRouter benchmark exports84.5%
820Gemma 4 31B ITOpenRouter harness, google/gemma-4-31b-it-20260402, Modal, public 90-day aggregate· OpenRouter benchmark exports84.4%
821DeepSeek V4 FlashOpenRouter harness, deepseek/deepseek-v4-flash-20260423, AtlasCloud, public 90-day aggregate· OpenRouter benchmark exports84.4%
822DeepSeek V4 Flash 0731OpenRouter harness, deepseek/deepseek-v4-flash-20260731, Chutes, public 90-day aggregate· OpenRouter benchmark exports84.4%
823Gemini 3.5 Flash LiteOpenRouter harness, google/gemini-3.5-flash-lite-20260721, auto-routing, public 90-day aggregate· OpenRouter benchmark exports84.4%
824GLM 5.2OpenRouter harness, z-ai/glm-5.2-20260616, Together, public 90-day aggregate· OpenRouter benchmark exports84.4%
825Step 3.5 FlashStep 3.5 technical report · Tables 12–13 · Python tool· Reported by StepFun· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.84.4%
826GLM 5.3OpenRouter harness, z-ai/glm-5.3-20260816, SiliconFlow, public 90-day aggregate· OpenRouter benchmark exports84.4%
827Inkling FP4OpenRouter harness, thinkingmachines/inkling-20260715, Together, public 90-day aggregate· OpenRouter benchmark exports84.4%
828GLM 5.3OpenRouter harness, z-ai/glm-5.3-20260816, PrimeIntellect, public 90-day aggregate· OpenRouter benchmark exports84.4%
829GLM 5.3OpenRouter harness, z-ai/glm-5.3-20260816, Relace, public 90-day aggregate· OpenRouter benchmark exports84.3%
830GLM 5.2OpenRouter harness, z-ai/glm-5.2-20260616, AtlasCloud, public 90-day aggregate· OpenRouter benchmark exports84.3%
831Gemma 4 31B ITGemma 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.84.3%
832MiMo V2 Flashtechnical report Table 9 · final evaluation· Reported by Xiaomi MiMo· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.84.3%
833GLM 5.3 FlashOpenRouter harness, z-ai/glm-5.3-flash-20260826, StreamLake, public 90-day aggregate· OpenRouter benchmark exports84.3%
834GLM 5.3OpenRouter harness, z-ai/glm-5.3-20260816, Decart, public 90-day aggregate· OpenRouter benchmark exports84.2%
835GLM 5.3 FlashOpenRouter harness, z-ai/glm-5.3-flash-20260826, Together, public 90-day aggregate· OpenRouter benchmark exports84.2%
836Qwen3.5 35B A3B· OpenEvals Leaderboard data· Reported by the model's developer. The source relaying it did not verify it.
Values from 2 sources
84.2%
837GLM 5.3OpenRouter harness, z-ai/glm-5.3-20260816, Mistral, public 90-day aggregate· OpenRouter benchmark exports84.2%
838Qwen 3.8 27BOpenRouter harness, qwen/qwen3.8-27b-20260814, Cloudflare, public 90-day aggregate· OpenRouter benchmark exports84.2%
839GLM 5.2OpenRouter harness, z-ai/glm-5.2-20260616, Crusoe, public 90-day aggregate· OpenRouter benchmark exports84.2%
840GPT-5.4 miniOpenRouter harness, openai/gpt-5.4-mini-20260317, OpenAI, public 90-day aggregate· OpenRouter benchmark exports84.2%
841Kimi K2.5OpenRouter harness, moonshotai/kimi-k2.5-0127, Phala, public 90-day aggregate· OpenRouter benchmark exports84.2%
842MiniMax M2.5OpenRouter harness, minimax/minimax-m2.5-20260211, DigitalOcean, public 90-day aggregate· OpenRouter benchmark exports84.2%
843MiniMax M2.5OpenRouter harness, minimax/minimax-m2.5-20260211, Novita, public 90-day aggregate· OpenRouter benchmark exports84.2%
844MiniMax M2.5OpenRouter harness, minimax/minimax-m2.5-20260211, StreamLake, public 90-day aggregate· OpenRouter benchmark exports84.2%
845GLM 5.3OpenRouter harness, z-ai/glm-5.3-20260816, Makora, public 90-day aggregate· OpenRouter benchmark exports84.1%
846GLM 5.3 FlashOpenRouter harness, z-ai/glm-5.3-flash-20260826, AtlasCloud, public 90-day aggregate· OpenRouter benchmark exports84.1%
847Qwen3.6 27BOpenRouter harness, qwen/qwen3.6-27b-20260422, DeepInfra, public 90-day aggregate· OpenRouter benchmark exports84.1%
848Gemma 4 31B ITOpenRouter harness, google/gemma-4-31b-it-20260402, PrimeIntellect, public 90-day aggregate· OpenRouter benchmark exports84.1%
849Gemma 4 31B ITOpenRouter harness, google/gemma-4-31b-it-20260402, SambaNova, public 90-day aggregate· OpenRouter benchmark exports84.1%
850GLM 5.2OpenRouter harness, z-ai/glm-5.2-20260616, Cloudflare, public 90-day aggregate· OpenRouter benchmark exports84.1%

Showing 801–850 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)