# τ²-bench results → GDPval-AA

Calibration · https://benchgap.net/calibration/1492 · data retrieved 2026-10-29

GDPval-AA is estimated from τ²-bench results with a Hill curve fitted on 68 models measured on both: y = 0.0000 + (1.2000 − 0.0000)·x^5.38 / (1.17736^5.38 + x^5.38), R² = 0.57, cross-validated error 10.9 pp. It is used for 53 estimates.

| Estimated model | τ²-bench results | GDPval-AA | Source |
|---|---|---|---|
| [GLM-5-Turbo](https://benchgap.net/model/glm-5-turbo.md) | 98.5% | 33.2% | estimated ± 10.9 pp, low confidence |
| [Qwen 3.6 Max (preview)](https://benchgap.net/model/qwen3-6-max-preview.md) | 95.9% | 29.9% | estimated ± 10.9 pp, low confidence |
| [Qwen3.5 397B (Reasoning)](https://benchgap.net/model/qwen3-5-397b-reasoning.md) | 95.6% | 29.5% | estimated ± 10.9 pp, low confidence |
| [Grok 4.1 Fast (Reasoning)](https://benchgap.net/model/grok-4-1-fast-reasoning.md) | 93.3% | 26.7% | estimated ± 10.9 pp, low confidence |
| [Claude Opus 4.6 (Adaptive)](https://benchgap.net/model/claude-opus-4-6-thinking.md) | 92.1% | 25.3% | estimated ± 10.9 pp, low confidence |
| [Qwen3.5-4B](https://benchgap.net/model/qwen3-5-4b.md) | 92.1% | 25.3% | estimated ± 10.9 pp, low confidence |
| [MiMo-V2-Omni](https://benchgap.net/model/mimo-v2-omni.md) | 91.2% | 24.2% | estimated ± 10.9 pp, low confidence |
| [Claude Opus 4.5 Thinking](https://benchgap.net/model/claude-opus-4-5-thinking.md) | 89.5% | 22.3% | estimated ± 10.9 pp, low confidence |
| [GPT-5 (medium)](https://benchgap.net/model/gpt-5-medium.md) | 86.5% | 19.2% | estimated ± 10.9 pp, low confidence |
| [GPT-5.1-Codex-Max](https://benchgap.net/model/gpt-5-1-codex-max.md) | 83.0% | 15.9% | estimated ± 10.9 pp, low confidence |
| [o3](https://benchgap.net/model/o3.md) | 80.7% | 13.9% | estimated ± 10.9 pp, low confidence |
| [Ternary Bonsai 2 27B](https://benchgap.net/model/ternary-bonsai-2-27b.md) | 80.2% | 13.5% | estimated ± 10.9 pp, low confidence |
| [GLM-4.6](https://benchgap.net/model/glm-4-6.md) | 76.9% | 11.0% | estimated ± 10.9 pp, low confidence |
| [Grok Code Fast 1](https://benchgap.net/model/grok-code-fast-1.md) | 75.7% | 10.2% | estimated ± 10.9 pp, low confidence |
| [Claude 4 Opus Thinking](https://benchgap.net/model/claude-4-opus-thinking.md) | 73.4% | 8.7% | estimated ± 10.9 pp, low confidence |
| [Claude 4.1 Opus Thinking](https://benchgap.net/model/claude-4-1-opus-thinking.md) | 71.4% | 7.6% | estimated ± 10.9 pp, low confidence |
| [Grok 4 Fast (Reasoning)](https://benchgap.net/model/grok-4-fast-reasoning.md) | 65.8% | 5.0% | estimated ± 10.9 pp, low confidence |
| [o1](https://benchgap.net/model/o1.md) | 62.6% | 3.9% | estimated ± 10.9 pp, low confidence |
| [Kimi K2](https://benchgap.net/model/kimi-k2.md) | 61.1% | 3.4% | estimated ± 10.9 pp, low confidence |
| [o4-mini](https://benchgap.net/model/o4-mini.md) | 55.6% | 2.1% | estimated ± 10.9 pp, low confidence |
| [Nemotron 3.5 Bolt E4B BF16](https://benchgap.net/model/nemotron-3-5-bolt-e4b-bf16.md) | 55.1% | 2.0% | estimated ± 10.9 pp, low confidence |
| [Claude 3.7 Sonnet Thinking](https://benchgap.net/model/claude-3-7-sonnet-thinking.md) | 54.7% | 1.9% | estimated ± 10.9 pp, low confidence |
| [Nemotron 3.5 Bolt E4B](https://benchgap.net/model/nemotron-3-5-bolt-e4b.md) | 52.3% | 1.5% | estimated ± 10.9 pp, low confidence |
| [Claude 3.7 Sonnet](https://benchgap.net/model/claude-3-7-sonnet.md) | 50.0% | 1.2% | estimated ± 10.9 pp, low confidence |
| [GLM-4.5-Air](https://benchgap.net/model/glm-4-5-air.md) | 46.5% | 0.8% | estimated ± 10.9 pp, low confidence |
| [DeepSeek V3.1 (Reasoning)](https://benchgap.net/model/deepseek-v3-1-reasoning.md) | 37.4% | 0.2% | estimated ± 10.9 pp, low confidence |
| [DeepSeek-R1](https://benchgap.net/model/deepseek-r1.md) | 36.5% | 0.2% | estimated ± 10.9 pp, low confidence |
| [DeepSeek V3.1](https://benchgap.net/model/deepseek-v3-1.md) | 34.8% | 0.2% | estimated ± 10.9 pp, low confidence |
| [Solar Pro 2](https://benchgap.net/model/solar-pro-2.md) | 31.9% | 0.1% | estimated ± 10.9 pp, low confidence |
| [Mistral Large 2](https://benchgap.net/model/mistral-large-2.md) | 30.7% | 0.1% | estimated ± 10.9 pp, low confidence |
| [Gemini 2.0 Flash](https://benchgap.net/model/gemini-2-0-flash.md) | 29.5% | 0.1% | estimated ± 10.9 pp, low confidence |
| [o3-mini](https://benchgap.net/model/o3-mini.md) | 28.7% | 0.1% | estimated ± 10.9 pp, low confidence |
| [Devstral Small](https://benchgap.net/model/devstral-small.md) | 28.4% | 0.1% | estimated ± 10.9 pp, low confidence |
| [Magistral Small](https://benchgap.net/model/magistral-small.md) | 26.6% | 0.0% | estimated ± 10.9 pp, low confidence |
| [Mistral Medium 3](https://benchgap.net/model/mistral-medium-3.md) | 24.3% | 0.0% | estimated ± 10.9 pp, low confidence |
| [Magistral Medium](https://benchgap.net/model/magistral-medium.md) | 23.1% | 0.0% | estimated ± 10.9 pp, low confidence |
| [Llama 3.1 70B](https://benchgap.net/model/llama-3-1-70b.md) | 21.6% | 0.0% | estimated ± 10.9 pp, low confidence |
| [Qwen3-Omni-30B-A3B-Thinking](https://benchgap.net/model/qwen3-omni-30b-a3b-thinking.md) | 21.3% | 0.0% | estimated ± 10.9 pp, low confidence |
| [Claude 3 Haiku](https://benchgap.net/model/claude-3-haiku.md) | 21.1% | 0.0% | estimated ± 10.9 pp, low confidence |
| [Exaone 4.0 1.2B](https://benchgap.net/model/exaone-4-0-1-2b.md) | 20.5% | 0.0% | estimated ± 10.9 pp, low confidence |
| [Granite-4.0-H-1B](https://benchgap.net/model/granite-4-0-h-1b.md) | 19.6% | 0.0% | estimated ± 10.9 pp, low confidence |
| [Gemini 2.5 Flash-Lite](https://benchgap.net/model/gemini-2-5-flash-lite.md) | 19.0% | 0.0% | estimated ± 10.9 pp, low confidence |
| [Llama 3.1 405B](https://benchgap.net/model/llama-3-1-405b.md) | 19.0% | 0.0% | estimated ± 10.9 pp, low confidence |
| [Qwen3-Omni-30B-A3B-Instruct](https://benchgap.net/model/qwen3-omni-30b-a3b-instruct.md) | 16.4% | 0.0% | estimated ± 10.9 pp, low confidence |
| [Gemini 2.5 Flash](https://benchgap.net/model/gemini-2-5-flash.md) | 14.9% | 0.0% | estimated ± 10.9 pp, low confidence |
| [Granite-4.0-H-350M](https://benchgap.net/model/granite-4-0-h-350m.md) | 14.6% | 0.0% | estimated ± 10.9 pp, low confidence |
| [Nova Pro](https://benchgap.net/model/nova-pro.md) | 14.0% | 0.0% | estimated ± 10.9 pp, low confidence |
| [Granite-4.0-350M](https://benchgap.net/model/granite-4-0-350m.md) | 13.2% | 0.0% | estimated ± 10.9 pp, low confidence |
| [Nemotron Ultra 253B](https://benchgap.net/model/nemotron-ultra-253b.md) | 11.4% | 0.0% | estimated ± 10.9 pp, low confidence |
| [LFM2.5-VL-1.6B-Extract](https://benchgap.net/model/lfm2-5-vl-1-6b-extract.md) | 8.5% | 0.0% | estimated ± 10.9 pp, low confidence |
| [Exaone 4.0 32B](https://benchgap.net/model/exaone-4-0-32b.md) | 4.1% | 0.0% | estimated ± 10.9 pp, low confidence |
| [Phi-4](https://benchgap.net/model/phi-4.md) | 0.0% | 0.0% | estimated ± 10.9 pp, low confidence |
| [Phi-4-reasoning](https://benchgap.net/model/phi-4-reasoning.md) | 0.0% | 0.0% | estimated ± 10.9 pp, low confidence |

## Fitted on

| Model | τ²-bench results | GDPval-AA |
|---|---|---|
| [Claude 3.5 Haiku](https://benchgap.net/model/claude-3-5-haiku.md) | 24.6% | 0.0% |
| [Claude Fable 5](https://benchgap.net/model/claude-fable.md) | 98.5% | 55.5% |
| [Claude Opus 4.7 (Adaptive)](https://benchgap.net/model/claude-opus-4-7-adaptive.md) | 88.6% | 42.6% |
| [Claude Opus 4.8](https://benchgap.net/model/claude-opus-4-8.md) | 94.4% | 47.7% |
| [Command A+](https://benchgap.net/model/command-a-plus.md) | 85.0% | 0.7% |
| [DeepSeek V3](https://benchgap.net/model/deepseek-v3.md) | 22.8% | 0.0% |
| [DeepSeek V3 0324](https://benchgap.net/model/deepseek-v3-0324.md) | 47.1% | 0.0% |
| [DeepSeek V4 Pro 0813](https://benchgap.net/model/deepseek-v4-pro-0813.md) | 96.2% | 54.5% |
| [Gemini 2.5 Pro](https://benchgap.net/model/gemini-2-5-pro.md) | 54.1% | 0.0% |
| [Gemini 3.1 Pro](https://benchgap.net/model/gemini-3-1-pro.md) | 95.6% | 14.6% |
| [Gemini 3.5 Flash](https://benchgap.net/model/gemini-3-5-flash.md) | 95.3% | 42.2% |
| [Gemma 3 27B](https://benchgap.net/model/gemma-3-27b.md) | 10.5% | 0.0% |
| [Gemma 4 12B](https://benchgap.net/model/gemma-4-12b.md) | 36.3% | 0.0% |
| [Gemma 4 26B A4B](https://benchgap.net/model/gemma-4-26b-a4b.md) | 43.6% | 3.2% |
| [Gemma 4 31B](https://benchgap.net/model/gemma-4-31b.md) | 59.9% | 5.9% |
| [Gemma 4 E2B](https://benchgap.net/model/gemma-4-e2b.md) | 20.8% | 0.0% |
| [Gemma 4 E4B](https://benchgap.net/model/gemma-4-e4b.md) | 20.8% | 0.0% |
| [GLM-4.7](https://benchgap.net/model/glm-4-7.md) | 95.9% | 25.7% |
| [GLM-5.1](https://benchgap.net/model/glm-5-1.md) | 97.7% | 30.8% |
| [GLM-5.2](https://benchgap.net/model/glm-5-2.md) | 99.1% | 43.6% |
| [GPT-4.1 mini](https://benchgap.net/model/gpt-4-1-mini.md) | 52.9% | 0.0% |
| [GPT-4.1 nano](https://benchgap.net/model/gpt-4-1-nano.md) | 17.3% | 0.0% |
| [GPT-4o](https://benchgap.net/model/gpt-4o.md) | 25.1% | 0.0% |
| [GPT-5.1](https://benchgap.net/model/gpt-5-1.md) | 81.9% | 16.3% |
| [GPT-5.4](https://benchgap.net/model/gpt-5-4.md) | 98.9% | 37.3% |
| [GPT-5.4 mini](https://benchgap.net/model/gpt-5-4-mini.md) | 93.4% | 25.6% |
| [GPT-5.4 nano](https://benchgap.net/model/gpt-5-4-nano.md) | 92.5% | 22.4% |
| [GPT-5.5](https://benchgap.net/model/gpt-5-5.md) | 98.0% | 42.5% |
| [GPT-5.6 Sol](https://benchgap.net/model/gpt-5-6-sol.md) | 85.1% | 55.4% |
| [GPT-5.6 Terra](https://benchgap.net/model/gpt-5-6-terra.md) | 86.3% | 47.6% |
| [GPT-5 (high)](https://benchgap.net/model/gpt-5-high.md) | 84.8% | 20.9% |
| [GPT-OSS 120B](https://benchgap.net/model/gpt-oss-120b.md) | 65.8% | 5.4% |
| [GPT-OSS 20B](https://benchgap.net/model/gpt-oss-20b.md) | 60.2% | 0.0% |
| [Grok 4.3](https://benchgap.net/model/grok-4-3.md) | 97.7% | 29.2% |
| [K-Exaone](https://benchgap.net/model/k-exaone.md) | 74.3% | 0.0% |
| [Kimi K2.6](https://benchgap.net/model/kimi-2-6.md) | 95.9% | 26.9% |
| [Kimi K2.5](https://benchgap.net/model/kimi-k2-5.md) | 95.9% | 17.0% |
| [Kimi K2.5 (Reasoning)](https://benchgap.net/model/kimi-k2-5-reasoning.md) | 95.9% | 17.0% |
| [Kimi K2.7 Code](https://benchgap.net/model/kimi-k2-7-code.md) | 90.1% | 26.9% |
| [Ling 2.6 Flash](https://benchgap.net/model/ling-2-6-flash.md) | 86.0% | 0.0% |
| [Llama 3.3 70B](https://benchgap.net/model/llama-3-3-70b.md) | 26.6% | 0.0% |
| [Llama 4 Maverick](https://benchgap.net/model/llama-4-maverick.md) | 17.8% | 0.0% |
| [Llama 4 Scout](https://benchgap.net/model/llama-4-scout.md) | 15.5% | 0.0% |
| [MiMo-V2.5-Pro](https://benchgap.net/model/mimo-v2-5-pro.md) | 94.2% | 31.0% |
| [MiMo-V2-Flash](https://benchgap.net/model/mimo-v2-flash.md) | 83.9% | 6.4% |
| [MiniMax M2.7](https://benchgap.net/model/minimax-m2-7.md) | 84.8% | 25.5% |
| [MiniMax M3](https://benchgap.net/model/minimax-m3.md) | 88.9% | 37.1% |
| [Mistral Large 3](https://benchgap.net/model/mistral-large-3.md) | 24.6% | 0.0% |
| [Mistral Medium 3.5 128B](https://benchgap.net/model/mistral-medium-3-5-128b.md) | 94.2% | 13.0% |
| [Mistral Small 3.1](https://benchgap.net/model/mistral-small-3-1.md) | 25.1% | 0.0% |
| [Mistral Small 4](https://benchgap.net/model/mistral-small-4.md) | 41.2% | 0.0% |
| [Mistral Small 4 (Reasoning)](https://benchgap.net/model/mistral-small-4-reasoning.md) | 18.4% | 0.0% |
| [Muse Spark](https://benchgap.net/model/muse-spark.md) | 91.5% | 24.9% |
| [Nemotron 3 Nano 30B](https://benchgap.net/model/nemotron-3-nano-30b.md) | 25.4% | 0.0% |
| [Nemotron 3 Nano Omni 30B A3B](https://benchgap.net/model/nemotron-3-nano-omni-30b-a3b.md) | 45.3% | 0.0% |
| [Nemotron 3 Ultra](https://benchgap.net/model/nemotron-3-ultra.md) | 83.3% | 33.1% |
| [North Mini Code](https://benchgap.net/model/north-mini-code-1-0.md) | 37.4% | 0.0% |
| [Qwen3.5-122B-A10B](https://benchgap.net/model/qwen3-5-122b-a10b.md) | 93.6% | 15.7% |
| [Qwen3.6-27B](https://benchgap.net/model/qwen3-6-27b.md) | 94.2% | 24.3% |
| [Qwen3.6-35B-A3B](https://benchgap.net/model/qwen3-6-35b-a3b.md) | 95.3% | 19.6% |
| [Qwen3.6 Plus](https://benchgap.net/model/qwen3-6-plus.md) | 97.7% | 24.5% |
| [Qwen3.7 Max](https://benchgap.net/model/qwen3-7-max.md) | 94.7% | 31.4% |
| [Qwen3.7 Plus](https://benchgap.net/model/qwen3-7-plus.md) | 93.0% | 13.3% |
| [Solar Pro 3](https://benchgap.net/model/solar-pro-3.md) | 86.3% | 0.0% |
| [Step 3.7 Flash](https://benchgap.net/model/step-3-7-flash.md) | 98.5% | 25.8% |
| [Trinity-Large-Preview](https://benchgap.net/model/trinity-large-preview.md) | 90.1% | 0.0% |
| [Trinity-Large-Thinking](https://benchgap.net/model/trinity-large-thinking.md) | 90.1% | 0.0% |
| [Ultravox v0.6 Llama 3.3 70B](https://benchgap.net/model/ultravox-v0-6-llama-3-3-70b.md) | 26.6% | 0.0% |