RuBench (Fresh-11): leaderboard
The Fresh-11 slice of RuBench (2026): the 11 Russian-specified repository fix tasks whose fix commits postdate every evaluated model cutoff, so all configurations run the same tasks. Pass@1 over three runs in percent, higher is better.
Metric: Pass@1 (%; the 11 tasks with fix commits after 2026-06-17, eligible for every configuration; pass@1, mean share of tasks resolved over three independent runs, judged by the withheld upstream maintainer regression tests; 25 repository-level fix tasks from aiohttp, aiogram, Laravel, NestJS and Fastify with task statements natively written in Russian; deployed agent CLI, model and reasoning effort as one configuration, product defaults otherwise, network access allowed). Source: arxiv.org. Saturation forecast: Around February 2027. 10 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.6 Luna (xHigh) | 75.8 | #121 (GPT-5.6 Luna) |
| 2 | Claude Opus 4.8 (xHigh) | 75.8 | #56 (Claude Opus 4.8) |
| 3 | Claude Sonnet 5 (xHigh) | 75.8 | #111 (Claude Sonnet 5) |
| 4 | GPT-5.5 (xHigh) | 72.7 | #49 (GPT-5.5) |
| 5 | Gemini 3.1 Pro (Preview) (High) | 48.5 | #94 (Gemini 3.1 Pro (Preview)) |
| 6 | Gemini 3.5 Flash (High) | 33.3 | #101 (Gemini 3.5 Flash) |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=rubench-fresh-11 · How It Works · Data refreshed daily, snapshot 2026-11-13.