GPQA Diamond leaderboard (Sep 2026): AI model scores
As of 1 Sep 2026, the highest reported Alibaba score on GPQA Diamond is Qwen 3.8 Max (Alibaba) at 93.69%, #6 on that leaderboard, per Vals AI. The source no longer updates this leaderboard.
What it measures
Graduate-level “Google-proof” science questions that require deep reasoning.
Graduate-level biology, physics and chemistry questions written to be “Google-proof”.
What the score means
Accuracy
Share of multiple-choice questions answered correctly.
Results from published AI leaderboards. Compare models within each benchmark; scores from different tests aren’t directly comparable. How we collect these results
Scores by model
Run by Vals AI. Leaderboard dated 1 Sep 2026 (no longer updated by the source). Last checked 22 Oct 2026. Showing top 25 of 138 models listed.
Highlight a provider: Anthropic OpenAI Google xAI Meta DeepSeek Alibaba Moonshot MiniMax
Highlighting Alibaba models. Other providers are dimmed. Show all providers
- Anthropic
- OpenAI
- xAI
- Meta
- DeepSeek
- Alibaba
- Moonshot
- MiniMax
- Accuracy
- Share of multiple-choice questions answered correctly.
All scores
| Rank | Model | Provider | Accuracy | Price in / out ($/1M tokens) |
|---|---|---|---|---|
| 1 | Gemini 3.1 Pro Preview (02/26) | 95.45% | $2 / $12 | |
| 2 | GPT-5.6 Sol | OpenAI | 95.20% | $4 / $20 |
| 3 | Grok 4.6 | xAI | 94.70% | $2 / $6 |
| 4 | Gemini 3.8 Flash | 94.44% | $1.5 / $7.5 | |
| 5 | Gemini 3.7 Flash | 93.94% | $1.5 / $7.5 | |
| 6 | Qwen 3.8 Max | Alibaba | 93.69% | $2 / $6 |
| 7 | Gemini 3.6 Flash | 93.43% | $1.5 / $7.5 | |
| 8 | Claude Opus 5 | Anthropic | 93.43% | $5 / $25 |
| 9 | Claude Fable 5.1 | Anthropic | 93.43% | $10 / $50 |
| 10 | Claude Fable 5 | Anthropic | 93.18% | $10 / $50 |
Show 15 more rows
| Rank | Model | Provider | Accuracy | Price in / out ($/1M tokens) |
|---|---|---|---|---|
| 11 | GPT 5.5 | OpenAI | 93.18% | $5 / $30 |
| 12 | Kimi K3 | Moonshot | 92.93% | $3 / $15 |
| 13 | Grok 4.5 | xAI | 92.93% | $2 / $6 |
| 14 | Gemini 3.5 Flash | 92.68% | $1.5 / $9 | |
| 15 | MiniMax-M3 | MiniMax | 92.68% | $0.6 / $2.4 |
| 16 | Claude Opus 4.8 | Anthropic | 92.42% | $5 / $25 |
| 17 | DeepSeek V4 Pro 0813 | DeepSeek | 92.42% | $1.32 / $3.96 |
| 18 | GPT-5.6 Luna | OpenAI | 91.67% | $0.2 / $1.2 |
| 19 | Gemini 3 Pro (11/25) | 91.67% | $2 / $12 | |
| 20 | GPT 5.2 | OpenAI | 91.67% | $1.75 / $14 |
| 21 | GPT 5.4 (xhigh) | OpenAI | 91.67% | $2.5 / $15 |
| 22 | Grok 4.3 | xAI | 91.41% | $1.25 / $2.5 |
| 23 | Muse Spark 1.1 | Meta | 91.16% | $1.25 / $4.25 |
| 24 | GPT-5.6 Terra | OpenAI | 90.91% | $2 / $12 |
| 25 | Qwen 3.7 Max | Alibaba | 90.15% | $2.5 / $7.5 |
Source: Vals AI leaderboard ↗
Best-ranked Alibaba model here: Qwen 3.8 Max (Alibaba). Browse prompts for any model →