TensorBench - IR: leaderboard
Metric: Pass rate (%) on the 8 intermediate-representation tasks (new CIN nodes, LLIR SSA form, iteration semantics, a codegen-architecture refactor); 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 2034. 7 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Opus 4.7 (xHigh) | 62.5 |
| 2 | Claude Opus 4.6 | 50 |
| 3 | GPT-5.5 (xHigh) | 37.5 |
| 4 | Qwen 3 Coder 480B A35B Instruct | 37.5 |
| 5 | GPT-5.4 (xHigh) | 12.5 |
| 6 | GPT-5.3 Codex (xHigh) | 12.5 |
| 7 | Gemini 3.1 Pro (Preview) | 0 |
Interactive version: theaggregate.ai/benchmark?slug=tensorbench-ir · How It Works · Data refreshed daily, snapshot 2026-09-29.