TensorBench - Runtime: leaderboard

Metric: Pass rate (%) on the 28 runtime tasks (caching and dispatch, memory management, autotuning, work stealing); each coding agent (a published CLI scaffold with its model) edits the open-source Scorch compiler-based tensor framework inside a per-task container, one run per task; a task passes when every test of the post-patch suite (pre-existing randomized regression tests plus the tests the agent adds) succeeds; higher is better. Source: arxiv.org. Saturation forecast: Around 2032. 7 models tracked.

Top models

#ModelScore
1GPT-5.5 (xHigh)60.7
2Qwen 3 Coder 480B A35B Instruct42.9
3Claude Opus 4.7 (xHigh)42.9
4Gemini 3.1 Pro (Preview)39.3
5Claude Opus 4.639.3
6GPT-5.4 (xHigh)35.7
7GPT-5.3 Codex (xHigh)25

Interactive version: theaggregate.ai/benchmark?slug=tensorbench-runtime · How It Works · Data refreshed daily, snapshot 2026-09-29.