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
251DeepSeek V4.1 Flash 1017OpenRouter harness, deepseek/deepseek-v4.1-flash-20261017, PrimaLabs, public 90-day aggregate· OpenRouter benchmark exports90.8%
252DeepSeek V4.1 Flash 1017OpenRouter harness, deepseek/deepseek-v4.1-flash-20261017, InferenceNet, public 90-day aggregate· OpenRouter benchmark exports90.8%
253DeepSeek V4.1 FlashOpenRouter harness, deepseek/deepseek-v4.1-flash-20260910, Alibaba, public 90-day aggregate· OpenRouter benchmark exports90.8%
254MiniMax M3OpenRouter harness, minimax/minimax-m3-20260531, AtlasCloud, public 90-day aggregate· OpenRouter benchmark exports90.8%
255DeepSeek V4 Flash 0731Qwen3.8 Omni Flash 官方发布表· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.90.8%
256Gemini 3 Pro PreviewSenseNova 6.7 Flash-Lite official benchmark chart · chart setting· Reported by SenseTime / OpenSenseNova· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.90.8%
257DeepSeek V4.1 FlashOpenRouter harness, deepseek/deepseek-v4.1-flash-20260910, Baidu, public 90-day aggregate· OpenRouter benchmark exports90.8%
258Kimi K2.6· Epoch AI Benchmarking Hub data90.8%
259DeepSeek V4.1 FlashOpenRouter harness, deepseek/deepseek-v4.1-flash-20260910, Mancer 2, public 90-day aggregate· OpenRouter benchmark exports90.8%
260MiniMax M3OpenRouter harness, minimax/minimax-m3-20260531, CoreWeave, public 90-day aggregate· OpenRouter benchmark exports90.8%
261DeepSeek V4.1 Flash 1017OpenRouter harness, deepseek/deepseek-v4.1-flash-20261017, Nebius, public 90-day aggregate· OpenRouter benchmark exports90.8%
262MiniMax M3OpenRouter harness, minimax/minimax-m3-20260531, Venice, public 90-day aggregate· OpenRouter benchmark exports90.8%
263DeepSeek V4.1 Flash 1017OpenRouter harness, deepseek/deepseek-v4.1-flash-20261017, Cohere, public 90-day aggregate· OpenRouter benchmark exports90.7%
264Inkling 2 SmallOpenRouter harness, thinkingmachines/inkling-2-small-20261107, auto-routing, public 90-day aggregate· OpenRouter benchmark exports90.7%
265MiniMax M3OpenRouter harness, minimax/minimax-m3-20260531, auto-routing, public 90-day aggregate· OpenRouter benchmark exports90.7%
266Hy4 previewOpenRouter harness, tencent/hy4-preview-20260827, DeepInfra, public 90-day aggregate· OpenRouter benchmark exports90.7%
267DeepSeek V4.1 Flash 1017OpenRouter harness, deepseek/deepseek-v4.1-flash-20261017, Modal, public 90-day aggregate· OpenRouter benchmark exports90.7%
268Inkling 2 SmallOpenRouter harness, thinkingmachines/inkling-2-small-20261107, PrimeIntellect, public 90-day aggregate· OpenRouter benchmark exports90.7%
269DeepSeek V4 ProOpenRouter harness, deepseek/deepseek-v4-pro-20260423, Novita, public 90-day aggregate· OpenRouter benchmark exports90.7%
270GPT 5.5low· Epoch AI Benchmarking Hub data90.7%
271DeepSeek V4 Flash 0731OpenRouter harness, deepseek/deepseek-v4-flash-20260731, Alibaba, public 90-day aggregate· OpenRouter benchmark exports90.7%
272DeepSeek V4.1 Flash 1017OpenRouter harness, deepseek/deepseek-v4.1-flash-20261017, Fireworks, public 90-day aggregate· OpenRouter benchmark exports90.6%
273DeepSeek V4.1 Flash 1017OpenRouter harness, deepseek/deepseek-v4.1-flash-20261017, Morph, public 90-day aggregate· OpenRouter benchmark exports90.6%
274MiMo V2.6 ProOpenRouter harness, xiaomi/mimo-v2.6-pro-20260921, Novita, public 90-day aggregate· OpenRouter benchmark exports90.6%
275GLM 5.2OpenRouter harness, z-ai/glm-5.2-20260616, DeepInfra, public 90-day aggregate· OpenRouter benchmark exports90.6%
276Claude Opus 5.5max· Epoch AI Benchmarking Hub data90.6%
277DeepSeek V4.1 Flash 1017OpenRouter harness, deepseek/deepseek-v4.1-flash-20261017, Decart, public 90-day aggregate· OpenRouter benchmark exports90.6%
278DeepSeek V4 Pro 0813OpenRouter harness, deepseek/deepseek-v4-pro-20260813, GMICloud, public 90-day aggregate· OpenRouter benchmark exports90.6%
279DeepSeek V4.1 FlashOpenRouter harness, deepseek/deepseek-v4.1-flash-20260910, SiliconFlow, public 90-day aggregate· OpenRouter benchmark exports90.6%
280DeepSeek V4.1 Flash 1017OpenRouter harness, deepseek/deepseek-v4.1-flash-20261017, Wafer, public 90-day aggregate· OpenRouter benchmark exports90.6%
281DeepSeek V4.1 Flash 1017OpenRouter harness, deepseek/deepseek-v4.1-flash-20261017, Relace, public 90-day aggregate· OpenRouter benchmark exports90.6%
282DeepSeek V4.1 Flash 1017OpenRouter harness, deepseek/deepseek-v4.1-flash-20261017, Makora, public 90-day aggregate· OpenRouter benchmark exports90.5%
283DeepSeek V4 ProOpenRouter harness, deepseek/deepseek-v4-pro-20260423, GMICloud, public 90-day aggregate· OpenRouter benchmark exports90.5%
284Hy4 previewOpenRouter harness, tencent/hy4-preview-20260827, InferenceNet, public 90-day aggregate· OpenRouter benchmark exports90.5%
285DeepSeek V4.1 FlashOpenRouter harness, deepseek/deepseek-v4.1-flash-20260910, StreamLake, public 90-day aggregate· OpenRouter benchmark exports90.5%
286Claude Opus 4.632k· Epoch AI Benchmarking Hub data90.5%
287Claude Sonnet 5xhigh· Epoch AI Benchmarking Hub data90.5%
288DeepSeek V4.1 Flash 1017OpenRouter harness, deepseek/deepseek-v4.1-flash-20261017, Sail Research, public 90-day aggregate· OpenRouter benchmark exports90.5%
289Kimi K2.6OpenRouter harness, moonshotai/kimi-k2.6-20260420, Crusoe, public 90-day aggregate· OpenRouter benchmark exports90.5%
290MiniMax M3OpenRouter harness, minimax/minimax-m3-20260531, ModelRun, public 90-day aggregate· OpenRouter benchmark exports90.5%
291Kimi K2.6Kimi K2.6 official benchmark table · thinking · temp 1.0 · top_p 1.0 · 262,144 context· Reported by Moonshot AI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.90.5%
292Qwen3.8 FlashOpenRouter harness, qwen/qwen3.8-flash-20260826, Alibaba, public 90-day aggregate· OpenRouter benchmark exports90.5%
293Muse Spark OpenOpenRouter harness, meta/muse-spark-open-20261015, Makora, public 90-day aggregate· OpenRouter benchmark exports90.5%
294DeepSeek V4 Flash 0731OpenRouter harness, deepseek/deepseek-v4-flash-20260731, DeepInfra, public 90-day aggregate· OpenRouter benchmark exports90.5%
295GPT-6 Lunamax· Epoch AI Benchmarking Hub data90.5%
296DeepSeek V4.1 Flash 1017OpenRouter harness, deepseek/deepseek-v4.1-flash-20261017, AtlasCloud, public 90-day aggregate· OpenRouter benchmark exports90.5%
297DeepSeek V4.1 FlashOpenRouter harness, deepseek/deepseek-v4.1-flash-20260910, DeepInfra, public 90-day aggregate· OpenRouter benchmark exports90.4%
298DeepSeek V4 Pro 0813OpenRouter harness, deepseek/deepseek-v4-pro-20260813, Relace, public 90-day aggregate· OpenRouter benchmark exports90.4%
299Gemini 3 FlashOpenRouter harness, google/gemini-3-flash-preview-20251217, Google, public 90-day aggregate· OpenRouter benchmark exports90.4%
300DeepSeek V4.1 Flash 1017OpenRouter harness, deepseek/deepseek-v4.1-flash-20261017, Mistral, public 90-day aggregate· OpenRouter benchmark exports90.4%

Showing 251–300 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)