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
301DeepSeek V4 Flash 0731OpenRouter harness, deepseek/deepseek-v4-flash-20260731, Novita, public 90-day aggregate· OpenRouter benchmark exports90.4%
302Gemini 3 Flashno tools · pass@1· Reported by Google DeepMind· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.90.4%
303Hy3 (Deep Infra)Hy3 official Benchmark Appendix · highest reasoning tier unless noted· Reported by Tencent Hunyuan· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.90.4%
304Qwen3.6 PlusQwen official release table · thinking mode unless benchmark requires otherwise· Reported by Alibaba Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.90.4%
305Grok 4.6OpenRouter harness, x-ai/grok-4.6-20260810, Amazon Bedrock, public 90-day aggregate· OpenRouter benchmark exports90.4%
306DeepSeek V4.1 Flash 1017OpenRouter harness, deepseek/deepseek-v4.1-flash-20261017, PrimeIntellect, public 90-day aggregate· OpenRouter benchmark exports90.3%
307DeepSeek V4.1 Flash 1017OpenRouter harness, deepseek/deepseek-v4.1-flash-20261017, Particle AI, public 90-day aggregate· OpenRouter benchmark exports90.3%
308GPT 5.4OpenRouter harness, openai/gpt-5.4-20260305, OpenAI, public 90-day aggregate· OpenRouter benchmark exports90.3%
309Qwen3.7 PlusQwen3.7-Plus release · recommended reasoning-effort=xhigh system prompt· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.90.3%
310Qwen3.7 PlusQwen3.8 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.90.3%
311DeepSeek V4.1 Flash 1017OpenRouter harness, deepseek/deepseek-v4.1-flash-20261017, Together, public 90-day aggregate· OpenRouter benchmark exports90.3%
312DeepSeek V4 Flash 0731OpenRouter harness, deepseek/deepseek-v4-flash-20260731, GMICloud, public 90-day aggregate· OpenRouter benchmark exports90.3%
313DeepSeek V4.1 Flash 1017OpenRouter harness, deepseek/deepseek-v4.1-flash-20261017, Parasail, public 90-day aggregate· OpenRouter benchmark exports90.3%
314DeepSeek V4.1 FlashOpenRouter harness, deepseek/deepseek-v4.1-flash-20260910, Novita, public 90-day aggregate· OpenRouter benchmark exports90.3%
315DeepSeek V4 Flash 0731OpenRouter harness, deepseek/deepseek-v4-flash-20260731, Baidu, public 90-day aggregate· OpenRouter benchmark exports90.3%
316Kimi K2.5OpenRouter harness, moonshotai/kimi-k2.5-0127, SiliconFlow, public 90-day aggregate· OpenRouter benchmark exports90.3%
317GPT 5.4OpenRouter harness, openai/gpt-5.4-20260305, auto-routing, public 90-day aggregate· OpenRouter benchmark exports90.2%
318Inkling 2 SmallOpenRouter harness, thinkingmachines/inkling-2-small-20261107, DeepInfra, public 90-day aggregate· OpenRouter benchmark exports90.2%
319DeepSeek V4 Pro 0813OpenRouter harness, deepseek/deepseek-v4-pro-20260813, BaseTen, public 90-day aggregate· OpenRouter benchmark exports90.2%
320MiniMax M3OpenRouter harness, minimax/minimax-m3-20260531, Parasail, public 90-day aggregate· OpenRouter benchmark exports90.2%
321Claude Opus 5.5OpenRouter harness, anthropic/claude-opus-5.5-20260921, auto-routing, public 90-day aggregate· OpenRouter benchmark exports90.2%
322Kimi K2.6OpenRouter harness, moonshotai/kimi-k2.6-20260420, Inceptron, public 90-day aggregate· OpenRouter benchmark exports90.2%
323Kimi K2.6OpenRouter harness, moonshotai/kimi-k2.6-20260420, Cloudflare, public 90-day aggregate· OpenRouter benchmark exports90.2%
324Claude Opus 4.7xhigh· Epoch AI Benchmarking Hub data90.2%
325GLM 5.3 Flashmax· Epoch AI Benchmarking Hub data90.2%
326Muse Spark OpenOpenRouter harness, meta/muse-spark-open-20261015, auto-routing, public 90-day aggregate· OpenRouter benchmark exports90.2%
327GPT 5.4OpenRouter harness, openai/gpt-5.4-20260305, Amazon Bedrock, public 90-day aggregate· OpenRouter benchmark exports90.2%
328Muse Spark OpenOpenRouter harness, meta/muse-spark-open-20261015, BaseTen, public 90-day aggregate· OpenRouter benchmark exports90.2%
329DeepSeek V4 Flash 0731OpenRouter harness, deepseek/deepseek-v4-flash-20260731, PrimeIntellect, public 90-day aggregate· OpenRouter benchmark exports90.1%
330DeepSeek V4.1 Flash 1017OpenRouter harness, deepseek/deepseek-v4.1-flash-20261017, BaseTen, public 90-day aggregate· OpenRouter benchmark exports90.1%
331MiMo V2.6 ProOpenRouter harness, xiaomi/mimo-v2.6-pro-20260921, Xiaomi, public 90-day aggregate· OpenRouter benchmark exports90.1%
332Claude Opus 5.5OpenRouter harness, anthropic/claude-opus-5.5-20260921, Anthropic, public 90-day aggregate· OpenRouter benchmark exports90.1%
333DeepSeek V4 FlashOpenRouter harness, deepseek/deepseek-v4-flash-20260423, Crusoe, public 90-day aggregate· OpenRouter benchmark exports90.1%
334Claude Opus 4.8OpenRouter harness, anthropic/claude-4.8-opus-20260528, Claude Platform on AWS, public 90-day aggregate· OpenRouter benchmark exports90.1%
335Claude Opus 4.8OpenRouter harness, anthropic/claude-4.8-opus-20260528, Google, public 90-day aggregate· OpenRouter benchmark exports90.1%
336DeepSeek V4.1 Flash 1017OpenRouter harness, deepseek/deepseek-v4.1-flash-20261017, auto-routing, public 90-day aggregate· OpenRouter benchmark exports90.1%
337DeepSeek V4.1 Flash 1017OpenRouter harness, deepseek/deepseek-v4.1-flash-20261017, Cloudflare, public 90-day aggregate· OpenRouter benchmark exports90.1%
338Claude Opus 4.8OpenRouter harness, anthropic/claude-4.8-opus-20260528, Azure, public 90-day aggregate· OpenRouter benchmark exports90.1%
339Claude Opus 4.8OpenRouter harness, anthropic/claude-4.8-opus-20260528, Amazon Bedrock, public 90-day aggregate· OpenRouter benchmark exports90.1%
340Claude Opus 5.5OpenRouter harness, anthropic/claude-opus-5.5-20260921, Amazon Bedrock, public 90-day aggregate· OpenRouter benchmark exports90%
341Claude Opus 5OpenRouter harness, anthropic/claude-opus-5-20260723, Azure, public 90-day aggregate· OpenRouter benchmark exports90%
342DeepSeek V4 Pro 0813OpenRouter harness, deepseek/deepseek-v4-pro-20260813, Sail Research, public 90-day aggregate· OpenRouter benchmark exports90%
343Kimi K2.6OpenRouter harness, moonshotai/kimi-k2.6-20260420, Phala, public 90-day aggregate· OpenRouter benchmark exports90%
344Claude Opus 4.8OpenRouter harness, anthropic/claude-4.8-opus-20260528, auto-routing, public 90-day aggregate· OpenRouter benchmark exports90%
345Muse Spark OpenOpenRouter harness, meta/muse-spark-open-20261015, Fireworks, public 90-day aggregate· OpenRouter benchmark exports90%
346Hy4 previewOpenRouter harness, tencent/hy4-preview-20260827, Cohere, public 90-day aggregate· OpenRouter benchmark exports90%
347DeepSeek V4 Pro 0813OpenRouter harness, deepseek/deepseek-v4-pro-20260813, Parasail, public 90-day aggregate· OpenRouter benchmark exports90%
348GPT 5.4OpenRouter harness, openai/gpt-5.4-20260305, Azure, public 90-day aggregate· OpenRouter benchmark exports90%
349DeepSeek V4 Flash Vision ExpOpenRouter harness, deepseek/deepseek-v4-flash-vision-exp-20260821, Novita, public 90-day aggregate· OpenRouter benchmark exports89.9%
350DeepSeek V4 Flash 0731OpenRouter harness, deepseek/deepseek-v4-flash-20260731, NextBit, public 90-day aggregate· OpenRouter benchmark exports89.9%

Showing 301–350 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)