CRSBench
Cyber reasoning systems find and patch bugs in 124 C, C++ and Java projects with 315 vulnerabilities. A patch passes only if no ground-truth proof of vulnerability still triggers and functionality tests pass.
The GitHub repo was created on 2025-10-31, and the public release month is not stated. The dataset is gated behind a data use agreement.
- Level
- Expert Specialists with years of training, often over hours of work
- Scored
- Checked automatically A program checks the result against an answer key, unit tests or the expected final state.
- Answers
- Agent task Many steps in a terminal, browser, desktop, app, game or robot.
Claude Opus 4.6 leads with 82.3% among models released by Apr 2026
Best score per model from the benchmark's own leaderboard, read on 2 Nov 2026. Where a model ran with several settings or harnesses, we keep its best.
The newest model with a score here came out in Apr 2026, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:
- GPT-6 Astra
- Claude Opus 5.5
- GPT-6.1 Terra
- GPT-6.1 Sol
- and 16 more
- 1 Claude Opus 4.6 82.3%
- 2 GLM-5.1 68.6%
- 3
Gemini 3 Flash 56.9% - 4 GPT-5.4 Mini 54.9%
- 5 Claude Haiku 4.5 31.4%
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | Claude Opus 4.6 | 82.3% | Claude Code | Feb 2026 | Official leaderboard |
| 2 | GLM-5.1 | 68.6% | Opencode | Apr 2026 | Official leaderboard |
| 3 | Gemini 3 Flash | 56.9% | Gemini CLI | Dec 2025 | Official leaderboard |
| 4 | GPT-5.4 Mini | 54.9% | Codex | Mar 2026 | Official leaderboard |
| 5 | Claude Haiku 4.5 | 31.4% | Claude Code | Oct 2025 | Official leaderboard |
Labs report scores for 1 model in their cards. We plot every source in the charts below.
3 models have held number one since 2023
Lab claims count from the day the card came out, independent runs from the model’s release.
Scores over time
All the scores we have, one dot each, so a model tested at several settings appears more than once.
Labs report it only in system cards
We read the headline results of 479 launch posts from 51 labs, and none of them list this benchmark. Some system cards do, below.
One system card reports it
Besides launch posts, we read the system cards, model cards and technical reports labs publish with a release. They report far more, including safety and frontier-risk tests.
| Date | Document | Score | As the document writes it | Section |
|---|---|---|---|---|
| 27 Jul 2026 |  MAI-Cyber-1-Flash model card | 65.1 | CRSBench, printed 0.651 (POV=1200), metric not defined in the card (External Cyber Benchmark Results) | Cyber risk |
More in Cybersecurity
- ExploitGym In use 869 real-vulnerability exploit-development tasks (userspace, V8, Linux kernel). Success means capturing a flag through the intended vulnerability. GPT-6 Astra 42.4%
- ExploitBench In use 41 recent V8 JavaScript-engine N-day vulnerabilities. Progress measured by 16 grader-verified capability flags across five tiers. GPT-6 Astra 100%
- CWE-bench In use Agents audit a repository and write a patch for a security flaw, without being told where the bug is. Covers six programming languages. Gemini 4 Carbon 71%
- Real-world vulnerability discovery In use Agents search popular open-source codebases to rediscover about 1,200 recent confirmed vulnerabilities across many languages. Reports share discovered. Gemini 4 Argon 86%
- Artificial Analysis Cyber Index In use Agents do enterprise cyber defense tasks from three benchmarks, CWE-Bench-AA, DeepsecBench-AA and CyberGym-E2E-AA, combined into one success rate.
Mistral Large 4 (Le Chonk, open weights) 51% - Big Sleep Evaluation (Google) In use Agents search large real codebases such as Chrome and Safari for critical, hard to find vulnerabilities. Scored by pass@1 success rate, run without safety guardrails. Gemini 3.5 Flash Cyber 72%