# Vals GPQA Diamond → GPQA

Calibration · https://benchgap.net/calibration/1546 · data retrieved 2026-10-29

GPQA is estimated from Vals GPQA Diamond with a Hill curve fitted on 25 models measured on both: y = 0.8051 + (1.2000 − 0.8051)·x^6.00 / (1.07066^6.00 + x^6.00), R² = 0.73, cross-validated error 2.6 pp. It is used for 0 estimates.

## Fitted on

| Model | Vals GPQA Diamond | GPQA |
|---|---|---|
| [Claude Opus 4.8](https://benchgap.net/model/claude-opus-4-8.md) | 92.4% | 93.6% |
| [Claude Sonnet 4.6](https://benchgap.net/model/claude-sonnet-4-6.md) | 85.6% | 89.9% |
| [DeepSeek V4 Flash 0731](https://benchgap.net/model/deepseek-v4-flash-0731.md) | 89.9% | 88.1% |
| [DeepSeek V4 Pro 0813](https://benchgap.net/model/deepseek-v4-pro-0813.md) | 92.4% | 90.1% |
| [Gemini 3.5 Flash](https://benchgap.net/model/gemini-3-5-flash.md) | 92.7% | 92.2% |
| [GLM-4.7](https://benchgap.net/model/glm-4-7.md) | 80.1% | 85.7% |
| [GLM-5.2](https://benchgap.net/model/glm-5-2.md) | 85.6% | 91.2% |
| [GPT-5.4 mini](https://benchgap.net/model/gpt-5-4-mini.md) | 83.1% | 88.0% |
| [GPT-5.4 nano](https://benchgap.net/model/gpt-5-4-nano.md) | 77.5% | 82.8% |
| [GPT-5.5](https://benchgap.net/model/gpt-5-5.md) | 93.2% | 93.6% |
| [GPT-5.6 Luna](https://benchgap.net/model/gpt-5-6-luna.md) | 91.7% | 92.3% |
| [GPT-5.6 Sol](https://benchgap.net/model/gpt-5-6-sol.md) | 95.2% | 94.6% |
| [GPT-5.6 Terra](https://benchgap.net/model/gpt-5-6-terra.md) | 90.9% | 92.9% |
| [Grok 4.3](https://benchgap.net/model/grok-4-3.md) | 91.4% | 90.1% |
| [Inkling](https://benchgap.net/model/inkling.md) | 87.1% | 87.9% |
| [Inkling-Small](https://benchgap.net/model/inkling-small.md) | 83.6% | 89.5% |
| [Kimi K2.6](https://benchgap.net/model/kimi-2-6.md) | 89.1% | 90.5% |
| [Kimi K3](https://benchgap.net/model/kimi-k3.md) | 92.9% | 93.5% |
| [Ling 3.0 Flash](https://benchgap.net/model/ling-3-0-flash.md) | 84.8% | 85.0% |
| [MiMo-V2-Flash](https://benchgap.net/model/mimo-v2-flash.md) | 59.3% | 83.7% |
| [Nemotron 3 Ultra](https://benchgap.net/model/nemotron-3-ultra.md) | 86.1% | 87.0% |
| [Qwen3.6 Plus](https://benchgap.net/model/qwen3-6-plus.md) | 87.4% | 90.4% |
| [Qwen3.7 Max](https://benchgap.net/model/qwen3-7-max.md) | 90.2% | 92.4% |
| [Qwen3.8-27B](https://benchgap.net/model/qwen3-8-27b.md) | 88.9% | 89.2% |
| [Qwen3.8 Max](https://benchgap.net/model/qwen3-8-max.md) | 93.7% | 92.6% |