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
551Claude Opus 5low· Epoch AI Benchmarking Hub data87.9%
552GLM 5.2low· Epoch AI Benchmarking Hub data87.9%
553GPT 5.2medium· Epoch AI Benchmarking Hub data87.9%
554Kimi K2.7 Code· Epoch AI Benchmarking Hub data87.9%
555Qwen3.7 Plus· Epoch AI Benchmarking Hub data87.9%
556DeepSeek V4 Flash 0731OpenRouter harness, deepseek/deepseek-v4-flash-20260731, Venice, public 90-day aggregate· OpenRouter benchmark exports87.9%
557Kimi K2.5OpenRouter harness, moonshotai/kimi-k2.5-0127, Novita, public 90-day aggregate· OpenRouter benchmark exports87.8%
558DeepSeek V4 Flash 0731OpenRouter harness, deepseek/deepseek-v4-flash-20260731, Nebius, public 90-day aggregate· OpenRouter benchmark exports87.8%
559Nemotron 3.5 Super VLOpenRouter harness, nvidia/nemotron-3.5-super-vl-20261015, DeepInfra, public 90-day aggregate· OpenRouter benchmark exports87.8%
560GLM 5.3 FlashOpenRouter harness, z-ai/glm-5.3-flash-20260826, Wafer, public 90-day aggregate· OpenRouter benchmark exports87.8%
561Qwen3.5 397B A17BOpenRouter harness, qwen/qwen3.5-397b-a17b-20260216, DigitalOcean, public 90-day aggregate· OpenRouter benchmark exports87.8%
562GPT-6 LunaOpenRouter harness, openai/gpt-6-luna-20260922, Amazon Bedrock, public 90-day aggregate· OpenRouter benchmark exports87.8%
563GLM 5· Epoch AI Benchmarking Hub data
Values from 2 sources
87.8%
564Qwen3.6 27BQwen 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.87.8%
565MiniMax M3.1 FlashOpenRouter harness, minimax/minimax-m3.1-flash-20261019, GMICloud, public 90-day aggregate· OpenRouter benchmark exports87.8%
566DeepSeek V4 ProOpenRouter harness, deepseek/deepseek-v4-pro-20260423, Azure, public 90-day aggregate· OpenRouter benchmark exports87.8%
567DeepSeek V4 Pro 0813OpenRouter harness, deepseek/deepseek-v4-pro-20260813, Together, public 90-day aggregate· OpenRouter benchmark exports87.7%
568GLM 5.2OpenRouter harness, z-ai/glm-5.2-20260616, Reka, public 90-day aggregate· OpenRouter benchmark exports87.7%
569DeepSeek V4 FlashOpenRouter harness, deepseek/deepseek-v4-flash-20260423, Alibaba, public 90-day aggregate· OpenRouter benchmark exports87.7%
570GPT-6 LunaOpenRouter harness, openai/gpt-6-luna-20260922, Azure, public 90-day aggregate· OpenRouter benchmark exports87.7%
571Qwen3.5 397B A17BOpenRouter harness, qwen/qwen3.5-397b-a17b-20260216, GMICloud, public 90-day aggregate· OpenRouter benchmark exports87.7%
572DeepSeek V4.1 FlashOpenRouter harness, deepseek/deepseek-v4.1-flash-20260910, Parasail, public 90-day aggregate· OpenRouter benchmark exports87.7%
573DeepSeek V4.1 FlashOpenRouter harness, deepseek/deepseek-v4.1-flash-20260910, Relace, public 90-day aggregate· OpenRouter benchmark exports87.7%
574GLM 5.3 FlashOpenRouter harness, z-ai/glm-5.3-flash-20260826, Z.AI, public 90-day aggregate· OpenRouter benchmark exports87.6%
575GPT 5.1high· Epoch AI Benchmarking Hub data87.6%
576GPT-5.6 LunaOpenRouter harness, openai/gpt-5.6-luna-20260709, auto-routing, public 90-day aggregate· OpenRouter benchmark exports87.6%
577MiniMax M3.1 FlashOpenRouter harness, minimax/minimax-m3.1-flash-20261019, auto-routing, public 90-day aggregate· OpenRouter benchmark exports87.6%
578DeepSeek V4 Flash 0731OpenRouter harness, deepseek/deepseek-v4-flash-20260731, Together, public 90-day aggregate· OpenRouter benchmark exports87.6%
579Kimi K2.5setting not named by the source· Reported by Moonshot AI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.87.6%
580Kimi K2.5· Epoch AI Benchmarking Hub data
Values from 2 sources
87.6%
581DeepSeek V4.1 FlashOpenRouter harness, deepseek/deepseek-v4.1-flash-20260910, Decart, public 90-day aggregate· OpenRouter benchmark exports87.6%
582Nemotron 3.5 Super VLOpenRouter harness, nvidia/nemotron-3.5-super-vl-20261015, Nebius, public 90-day aggregate· OpenRouter benchmark exports87.6%
583Kimi K2.7 CodeOpenRouter harness, moonshotai/kimi-k2.7-code-20260612, Ambient, public 90-day aggregate· OpenRouter benchmark exports87.5%
584MiniMax M3.1 FlashOpenRouter harness, minimax/minimax-m3.1-flash-20261019, Makora, public 90-day aggregate· OpenRouter benchmark exports87.5%
585MiniMax M3.1 FlashOpenRouter harness, minimax/minimax-m3.1-flash-20261019, Together, public 90-day aggregate· OpenRouter benchmark exports87.5%
586Claude Sonnet 5OpenRouter harness, anthropic/claude-sonnet-5-20260630, Azure, public 90-day aggregate· OpenRouter benchmark exports87.5%
587Nemotron 3.5 Super VLOpenRouter harness, nvidia/nemotron-3.5-super-vl-20261015, Together, public 90-day aggregate· OpenRouter benchmark exports87.5%
588DeepSeek V4 ProOpenRouter harness, deepseek/deepseek-v4-pro-20260423, Reka, public 90-day aggregate· OpenRouter benchmark exports87.5%
589DeepSeek V4 FlashOpenRouter harness, deepseek/deepseek-v4-flash-20260423, DigitalOcean, public 90-day aggregate· OpenRouter benchmark exports87.5%
590DeepSeek V4 Flash 0731OpenRouter harness, deepseek/deepseek-v4-flash-20260731, Makora, public 90-day aggregate· OpenRouter benchmark exports87.5%
591DeepSeek V4 FlashOpenRouter harness, deepseek/deepseek-v4-flash-20260423, Azure, public 90-day aggregate· OpenRouter benchmark exports87.4%
592Qwen3.8 MaxOpenRouter harness, qwen/qwen3.8-2.4t-a95b-20260812, auto-routing, public 90-day aggregate· OpenRouter benchmark exports87.4%
593Claude Sonnet 5OpenRouter harness, anthropic/claude-sonnet-5-20260630, Amazon Bedrock, public 90-day aggregate· OpenRouter benchmark exports87.4%
594DeepSeek V4.1 FlashOpenRouter harness, deepseek/deepseek-v4.1-flash-20260910, BaseTen, public 90-day aggregate· OpenRouter benchmark exports87.4%
595Qwen3.5 397B A17BOpenRouter harness, qwen/qwen3.5-397b-a17b-20260216, Phala, public 90-day aggregate· OpenRouter benchmark exports87.4%
596DeepSeek V4.1 FlashOpenRouter harness, deepseek/deepseek-v4.1-flash-20260910, Makora, public 90-day aggregate· OpenRouter benchmark exports87.4%
597DeepSeek V4 FlashDeepSeek V4 official model card · Think High· Reported by DeepSeek· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.87.4%
598Claude Sonnet 4.632k· Epoch AI Benchmarking Hub data87.4%
599GPT-5.6 Terralow· Epoch AI Benchmarking Hub data87.4%
600Qwen3.6 Max Preview· Epoch AI Benchmarking Hub data87.4%

Showing 551–600 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)