Reasoning · Benchmark profile
Graduate-level Google-proof Q&A expansion beyond GPQA Diamond for harder knowledge-reasoning items.
Data verified 15 July 2026 · Methodology 1.2.0
Benchmark score on SuperGPQA
Claude Opus 4.6 leads at 95%, followed by Claude Sonnet 4.6 (95%) and Qwen3.7-Max (73.6%).
Visual analysis
Three views of the same sourced leaderboard: model placement, the score curve, and descriptive provider averages. These visuals never alter ranking calculations.
8 ranked models · higher is better
| Model | Rank | Provider | Score |
|---|---|---|---|
| Claude Opus 4.6 | #1 | Anthropic | 95% |
| Claude Sonnet 4.6 | #2 | Anthropic | 95% |
| Qwen3.7-Max | #3 | Alibaba Cloud | 73.6% |
| Qwen3.6 Plus | #4 | Alibaba Cloud | 71.6% |
| Qwen3.7-Plus | #5 | Alibaba Cloud | 71.4% |
| Claude Opus 4.5 | #6 | Anthropic | 70.6% |
| Kimi K2.5 | #7 | Moonshot AI | 69.2% |
| GLM-5 | #8 | Z.AI | 66.8% |
Shows how quickly scores separate down the field
| Model | Rank | Provider | Score |
|---|---|---|---|
| Claude Opus 4.6 | #1 | Anthropic | 95% |
| Claude Sonnet 4.6 | #2 | Anthropic | 95% |
| Qwen3.7-Max | #3 | Alibaba Cloud | 73.6% |
| Qwen3.6 Plus | #4 | Alibaba Cloud | 71.6% |
| Qwen3.7-Plus | #5 | Alibaba Cloud | 71.4% |
| Claude Opus 4.5 | #6 | Anthropic | 70.6% |
| Kimi K2.5 | #7 | Moonshot AI | 69.2% |
| GLM-5 | #8 | Z.AI | 66.8% |
Average within the visible cohort · descriptive only
| Provider | Models | Average score |
|---|---|---|
| Anthropic | 3 | 86.9% |
| Alibaba Cloud | 3 | 72.2% |
| Moonshot AI | 1 | 69.2% |
| Z.AI | 1 | 66.8% |
One best score per model · higher is better
| Rank | Model | Provider | License | Score |
|---|---|---|---|---|
| #1 | Claude Opus 4.6 claude-opus-4-6 | Anthropic | closed | 95% |
| #2 | Claude Sonnet 4.6 claude-sonnet-4-6 | Anthropic | closed | 95% |
| #3 | Qwen3.7-Max qwen3.7-max | Alibaba Cloud | — | 73.6% |
| #4 | Qwen3.6 Plus qwen3-6-plus | Alibaba Cloud | closed | 71.6% |
| #5 | Qwen3.7-Plus qwen3.7-plus | Alibaba Cloud | — | 71.4% |
| #6 | Claude Opus 4.5 claude-opus-4-5 | Anthropic | closed | 70.6% |
| #7 | Kimi K2.5 kimi-k2.5 | Moonshot AI | — | 69.2% |
| #8 | GLM-5 glm-5 | Z.AI | open | 66.8% |
The top of this snapshot is led by Claude Opus 4.6 at 95%; third place is 21.4 points behind. The top-8 spread is 28.2 points.
About SuperGPQA
Graduate-level Google-proof Q&A expansion beyond GPQA Diamond for harder knowledge-reasoning items. Results stay tied to the exact model variant and evaluation system. Multiple systems for the same model use the best published score on this page; overall Lumina scoring uses the median of ranking-eligible rows.
Open benchmark source ↗FAQ
Graduate-level Google-proof Q&A expansion beyond GPQA Diamond for harder knowledge-reasoning items.
Claude Opus 4.6 by Anthropic currently leads with 95%.
8 models in the LuminaBench cohort have a qualifying score on this benchmark.
Yes. This family is ranking-weighted at 2.5% of its capability category. Claude Opus 4.6 is overall #25.
Related
