GPQA Diamond Results Across LLMs
Vendor-reported GPQA Diamond results across showcased LLMs. Graduate-level science QA
What this table contains
Graduate-level science QA
15 models with results · 15 missing · 30 showcased models. Missing scores are not zero or failed tests. Headline selection is the default; source dates and setups can differ.
Interactive results & effort preference → · Results over time → · Methodology
| Model | Headline score | Source/record date | Evidence |
|---|---|---|---|
| GPT-6 AstraOpenAI | 96%unknown effort | 2026-09-03 | All 1 recorded result & sources96% · raw 96 % Headline · unknown effort · Own vendor Source/record date: 2026-09-03 https://openai.com/index/gpt-6-astra/ |
| Gemini 3.8 FlashGoogle | 95.3%unknown effort | 2026-09-03 | All 1 recorded result & sources95.3% · raw 95.3 % Headline · unknown effort · Peer vendor Source/record date: 2026-09-03 As reported by OpenAI https://openai.com/index/gpt-6-astra/ |
| Claude Fable 5.1Anthropic | 93.7%unknown effort | 2026-09-03 | All 1 recorded result & sources93.7% · raw 93.7 % Headline · unknown effort · Peer vendor Source/record date: 2026-09-03 As reported by OpenAI https://openai.com/index/gpt-6-astra/ |
| Kimi K3Moonshot AI | 93.5%max effort | 2026-07-16 | All 1 recorded result & sources93.5% · raw 93.5 % Headline · max effort · Own vendor Source/record date: 2026-07-16 GPQA Diamond; reasoning_effort=max, temp=1.0, top_p=0.95 https://github.com/MoonshotAI/Kimi-K3 |
| GPT-5.6 TerraOpenAI | 92.9%unknown effort | 2026-07-09 | All 1 recorded result & sources92.9% · raw 92.9 % Headline · unknown effort · Own vendor Source/record date: 2026-07-09 https://openai.com/index/gpt-5-6 |
| Qwen3.8 Max (0902)Alibaba (Qwen) | 92.6%unknown effort | 2026-09-02 | All 1 recorded result & sources92.6% · raw 92.6 % Headline · unknown effort · Third-party Source/record date: 2026-09-02 GPQA Diamond https://www.datacamp.com/blog/qwen3-8-max |
| Pareto 26.10 PreviewUnbiased | 92.4%unknown effort | 2026-10-01 | All 1 recorded result & sources92.4% · raw 92.4 % Headline · unknown effort · Own vendor Source/record date: 2026-10-01 Unbiased preliminary vendor result; mean task cost $0.004. Unbiased says results may change and denominator and cost methods need confirmation. https://unbiased.ai/blog/pareto-26-10-preview/ |
| Hy4 previewTencent | 92.3%unknown effort | 2026-08-28 | All 1 recorded result & sources92.3% · raw 92.3 % Headline · unknown effort · Own vendor Source/record date: 2026-08-28 Vendor chart transcription https://hy4.site/benchmarks/hy4-benchmarks |
| Qwen3.8 FlashAlibaba (Qwen) | 91.7%unknown effort | 2026-08-26 | All 1 recorded result & sources91.7% · raw 91.7 % Headline · unknown effort · Own vendor Source/record date: 2026-08-26 Qwen3.8-Flash-Next blog table (GPQA Diamond) https://qwen.ai/blog?id=qwen3.8-flash-next |
| Qwen3.8 Omni FlashAlibaba (Qwen) | 91%xhigh effort | 2026-09-18 | All 1 recorded result & sources91% · raw 91 % Headline · xhigh effort · Third-party Source/record date: 2026-09-18 GPQA Diamond — launch secondary transcription https://www.cometapi.com/what-is-qwen3-8-omni-flash-and-how-to-access/ |
| DeepSeek V4.1 FlashDeepSeek | 90.9%unknown effort | 2026-09-10 | All 1 recorded result & sources90.9% · raw 90.9 % Headline · unknown effort · Own vendor Source/record date: 2026-09-10 GPQA Diamond Pass@1 https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash |
| Qwen3.8 27BAlibaba (Qwen) | 89.2%unknown effort | 2026-08-14 | All 2 recorded results & sources89.2% · raw 89.2 % Headline · unknown effort · Own vendor Source/record date: 2026-08-14 GPQA Diamond https://huggingface.co/Qwen/Qwen3.8-27B89.2% · raw 89.2 % Alternative · unknown effort · Peer vendor Source/record date: 2026-10-03 Harness: Aleph Alpha eval-framework/Harbor, released post-trained checkpoint. Kolibri reasoning effort high; peers use vendor-documented sampling. Technical report Tables 27-29 (pp. 96-101). Chain-of-thought, avg@8. As reported by Aleph Alpha for Qwen3.8 27B; preserves existing own-vendor headline when present. https://huggingface.co/Aleph-Alpha/Kolibri-1 |
| GLM-5.3Z.ai | 88.1%unknown effort | 2026-09-10 | All 1 recorded result & sources88.1% · raw 88.1 % Headline · unknown effort · Peer vendor Source/record date: 2026-09-10 from chart/figure as reported on DeepSeek V4.1 Flash announce table (peer spillover) https://api-docs.deepseek.com/news/news260910 |
| Kolibri-1Aleph Alpha | 84.3%high effort | 2026-10-03 | All 1 recorded result & sources84.3% · raw 84.3 % Headline · high effort · Own vendor Source/record date: 2026-10-03 Harness: Aleph Alpha eval-framework/Harbor, released post-trained checkpoint. Kolibri reasoning effort high; peers use vendor-documented sampling. Technical report Tables 27-29 (pp. 96-101). Chain-of-thought, avg@8. https://huggingface.co/Aleph-Alpha/Kolibri-1 |
| Mercury 2.5Inception | 79%unknown effort | 2026-09-08 | All 1 recorded result & sources79% · raw 79 % Headline · unknown effort · Own vendor Source/record date: 2026-09-08 from chart/figure on Inception Mercury 2.5 release post (alt: Mercury 2.5 vs Mercury 2.0) https://www.inceptionlabs.ai/blog/introducing-mercury-2-5 |
| Claude Opus 5.5Anthropic | — | — | No recorded result |
| Claude Sonnet 5.5Anthropic | — | — | No recorded result |
| Gemini 4 ArgonGoogle | — | — | No recorded result |
| GLM-5.3-FlashZ.ai | — | — | No recorded result |
| GPT-6 LunaOpenAI | — | — | No recorded result |
| GPT-6.1 SolOpenAI | — | — | No recorded result |
| Grok 4.7xAI | — | — | No recorded result |
| Ling 3.0 Flash VLinclusionAI | — | — | No recorded result |
| Ling 3.1 FlashinclusionAI | — | — | No recorded result |
| MiMo-V2.6-Distill-Qwen-9BXiaomi | — | — | No recorded result |
| MiMo-V2.6-FlashXiaomi | — | — | No recorded result |
| MiMo-V2.6-ProXiaomi | — | — | No recorded result |
| Muse Spark 1.3Meta | — | — | No recorded result |
| Naive-N0.5-FlashNaiveAI | — | — | No recorded result |
| Nex-N2.5-ProNex AGI | — | — | No recorded result |
Benchmark source references
- Kolibri technical report · vendor_system_card
Use the fair-comparison guide before interpreting results from different configurations.