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
1401DeepSeek V3OpenRouter harness, deepseek/deepseek-chat-v3-0324, Crusoe, public 90-day aggregate· OpenRouter benchmark exports63.6%
1402Qwen3 30B A3B InstructOpenRouter harness, qwen/qwen3-30b-a3b-instruct-2507, CoreWeave, public 90-day aggregate· OpenRouter benchmark exports63.6%
1403GPT-5.6 Lunanone· Epoch AI Benchmarking Hub data63.6%
1404Qwen3 30B A3B InstructOpenRouter harness, qwen/qwen3-30b-a3b-instruct-2507, SiliconFlow, public 90-day aggregate· OpenRouter benchmark exports63.6%
1405Nemotron 3 Nano 30B A3BOpenRouter harness, nvidia/nemotron-3-nano-30b-a3b, Crusoe, public 90-day aggregate· OpenRouter benchmark exports63.5%
1406Nemotron 3.5 LightningOpenRouter harness, nvidia/nemotron-3.5-lightning-20260807, Fireworks, public 90-day aggregate· OpenRouter benchmark exports63.4%
1407Trinity Large PreviewTrinity technical report · Table 4 · exact reported precision· Reported by Arcee AI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.63.3%
1408Trinity Large PreviewTrinity-Large-Preview Model Card · rounded to one decimal· Reported by Arcee AI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.63.3%
1409Ling 3.0 Flash FinOpenRouter harness, inclusionai/ling-3.0-flash-fin-20260827, Novita, public 90-day aggregate· OpenRouter benchmark exports63.2%
1410Nemotron 3 Nano 30B A3BOpenRouter harness, nvidia/nemotron-3-nano-30b-a3b, auto-routing, public 90-day aggregate· OpenRouter benchmark exports63.1%
1411Qwen3 30B A3B InstructOpenRouter harness, qwen/qwen3-30b-a3b-instruct-2507, StreamLake, public 90-day aggregate· OpenRouter benchmark exports63.1%
1412Qwen3 30B A3B InstructOpenRouter harness, qwen/qwen3-30b-a3b-instruct-2507, DekaLLM, public 90-day aggregate· OpenRouter benchmark exports63.1%
1413GPT OSS 20bOpenRouter harness, openai/gpt-oss-20b, SiliconFlow, public 90-day aggregate· OpenRouter benchmark exports63%
1414Nemotron 3 Nano 30B A3BOpenRouter harness, nvidia/nemotron-3-nano-30b-a3b, Nebius, public 90-day aggregate· OpenRouter benchmark exports62.7%
1415Gemma 4 26B A4BOpenRouter harness, google/gemma-4-26b-a4b-it-20260403, Darkbloom, public 90-day aggregate· OpenRouter benchmark exports62.7%
1416DeepSeek V3.1OpenRouter harness, deepseek/deepseek-chat-v3.1, auto-routing, public 90-day aggregate· OpenRouter benchmark exports62.5%
1417MiMo V2.5 ProOpenRouter harness, xiaomi/mimo-v2.5-pro-20260422, AtlasCloud, public 90-day aggregate· OpenRouter benchmark exports62.5%
1418O1 Minihigh· Epoch AI Benchmarking Hub data62.4%
1419DeepSeek ChatOpenRouter harness, deepseek/deepseek-chat-v3, auto-routing, public 90-day aggregate· OpenRouter benchmark exports62.1%
1420Qwen3 30B A3B InstructOpenRouter harness, qwen/qwen3-30b-a3b-instruct-2507, PrimeIntellect, public 90-day aggregate· OpenRouter benchmark exports62.1%
1421Qwen3 32BOpenRouter harness, qwen/qwen3-32b-04-28, Nebius, public 90-day aggregate· OpenRouter benchmark exports62.1%
1422Qwen3 30B A3B InstructOpenRouter harness, qwen/qwen3-30b-a3b-instruct-2507, Nebius, public 90-day aggregate· OpenRouter benchmark exports62.1%
1423GLM 4.7 FlashOpenRouter harness, z-ai/glm-4.7-flash-20260119, Cloudflare, public 90-day aggregate· OpenRouter benchmark exports62%
1424Qwen3 4B Instruct 2507· OpenEvals Leaderboard data· Reported by the model's developer. The source relaying it did not verify it.
Values from 2 sources
62%
1425GPT OSS 120BOpenRouter harness, openai/gpt-oss-120b, Phala, public 90-day aggregate· OpenRouter benchmark exports62%
1426GPT OSS 20bOpenRouter harness, openai/gpt-oss-20b, Groq, public 90-day aggregate· OpenRouter benchmark exports62%
1427Qwen3 30B A3B· Epoch AI Benchmarking Hub data61.7%
1428Qwen3 CoderOpenRouter harness, qwen/qwen3-coder-480b-a35b-07-25, Novita, public 90-day aggregate· OpenRouter benchmark exports61.5%
1429Qwen3 14BOpenRouter harness, qwen/qwen3-14b-04-28, DeepInfra, public 90-day aggregate· OpenRouter benchmark exports61.2%
1430Qwen3 CoderOpenRouter harness, qwen/qwen3-coder-480b-a35b-07-25, Google, public 90-day aggregate· OpenRouter benchmark exports61%
1431Qwen3 235B A22B Instruct 2507OpenRouter harness, qwen/qwen3-235b-a22b-07-25, AtlasCloud, public 90-day aggregate· OpenRouter benchmark exports60.9%
1432Qwen3 30B A3BOpenRouter harness, qwen/qwen3-30b-a3b-04-28, DeepInfra, public 90-day aggregate· OpenRouter benchmark exports60.9%
1433GLM 4.7OpenRouter harness, z-ai/glm-4.7-20251222, Mancer 2, public 90-day aggregate· OpenRouter benchmark exports60.8%
1434GPT OSS 20bmedium· Epoch AI Benchmarking Hub data60.8%
1435qwen3-vl-30bOpenRouter harness, qwen/qwen3-vl-30b-a3b-instruct, SiliconFlow, public 90-day aggregate· OpenRouter benchmark exports60.8%
1436Qwen3 CoderOpenRouter harness, qwen/qwen3-coder-480b-a35b-07-25, auto-routing, public 90-day aggregate· OpenRouter benchmark exports60.7%
1437DeepSeek V3.2OpenRouter harness, deepseek/deepseek-v3.2-20251201, Mara, public 90-day aggregate· OpenRouter benchmark exports60.6%
1438Qwen3 CoderOpenRouter harness, qwen/qwen3-coder-480b-a35b-07-25, DeepInfra, public 90-day aggregate· OpenRouter benchmark exports60.6%
1439Claude Haiku 4.5· Epoch AI Benchmarking Hub data60.5%
1440qwen3-vl-30bOpenRouter harness, qwen/qwen3-vl-30b-a3b-instruct, DeepInfra, public 90-day aggregate· OpenRouter benchmark exports60.1%
1441Qwen3 CoderOpenRouter harness, qwen/qwen3-coder-480b-a35b-07-25, Venice, public 90-day aggregate· OpenRouter benchmark exports60%
1442Qwen3 CoderOpenRouter harness, qwen/qwen3-coder-480b-a35b-07-25, Alibaba, public 90-day aggregate· OpenRouter benchmark exports60%
1443GPT OSS 20bOpenRouter harness, openai/gpt-oss-20b, PrimeIntellect, public 90-day aggregate· OpenRouter benchmark exports59.9%
1444Qwen3 32BOpenRouter harness, qwen/qwen3-32b-04-28, auto-routing, public 90-day aggregate· OpenRouter benchmark exports59.9%
1445Qwen3 14BOpenRouter harness, qwen/qwen3-14b-04-28, NextBit, public 90-day aggregate· OpenRouter benchmark exports59.8%
1446Gemini 2.5 Flash-LiteOpenRouter harness, google/gemini-2.5-flash-lite, Google AI Studio, public 90-day aggregate· OpenRouter benchmark exports59.6%
1447Nemotron 3 Nano 30B A3BOpenRouter harness, nvidia/nemotron-3-nano-30b-a3b, Novita, public 90-day aggregate· OpenRouter benchmark exports59.6%
1448Mistral Medium 3· Epoch AI Benchmarking Hub data59.5%
1449O1 Minimedium· Epoch AI Benchmarking Hub data59.5%
1450DeepSeek V3.2OpenRouter harness, deepseek/deepseek-v3.2-20251201, SambaNova, public 90-day aggregate· OpenRouter benchmark exports59.3%

Showing 1,401–1,450 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)