Skip to main content
Radar

Five or fewer confirmed AI changes, with original sources, on mornings when something changed.A free source-linked morning brief.

Start free brief
BenchLM

CyberGym

A cybersecurity task benchmark for evaluating defensive cyber workflows and vulnerability-oriented agent performance.

Data verified 24 confirmed releases in the last 30 daysStart free brief

Benchmark score on CyberGym — August 22, 2026

We mirror the published score view for CyberGym. Fugu Cyber leads the public snapshot at 86.9%, followed by GPT-5.6 Sol (84.5%) and GLM-5.3 (84.5%). We do not use these results to rank models overall.

1Closed

Fugu Cyber

Sakana AI

86.9%
Context 1M
2Closed

GPT-5.6 Sol

OpenAI

84.5%
Overall 81.73Context 1.05M
3Closed

GLM-5.3

Z.AI

84.5%
Overall 62.53Context 1M
20 modelsAgenticCurrentDisplay onlyUpdated August 22, 2026

Benchmark score table (20 models)

Score
1
Fugu CyberSakana AI · Closed
86.9%
2
GPT-5.6 SolOpenAI · Closed
84.5%
3
GLM-5.3Z.AI · Closed
84.5%
4
Claude Mythos 5Anthropic · Closed
83.8%
5
DeepSeek V4 Pro 0813DeepSeek · Closed
83.3%
6
Gemini 3.5 Flash CyberGoogle · Closed
83.2%
7
Claude Mythos PreviewAnthropic · Closed
83.1%
8
GPT-5.6 TerraOpenAI · Closed
81.8%
9
GPT-5.5OpenAI · Closed
81.8%
10
GPT-5.4OpenAI · Closed
79.0%
11
GPT-5.6 LunaOpenAI · Closed
77.9%
12
DeepSeek V4 Flash 0731DeepSeek · Closed
76.7%
13
Claude Opus 4.7 (Adaptive)Anthropic · Closed
73.1%
14
GLM-5.1Z.AI · Open weight
68.7%
15
Claude Opus 4.6Anthropic · Closed
66.6%
16
Claude Sonnet 4.6Anthropic · Closed
65.2%
17
Muse Spark 1.1Meta · Closed
59.0%
18
Claude Opus 4.5Anthropic · Closed
50.6%
19
Muse SparkMeta · Closed
43.5%
20
GLM-5Z.AI · Open weight
43.2%

The published CyberGym snapshot places Fugu Cyber first at 86.9%. The third row is 2.4 points behind. The broader top-10 range is 7.9 points, so many of the published results sit in a relatively narrow band.

20 models have been evaluated on CyberGym. The benchmark falls in the Agentic category. This category carries a 22% weight in BenchLM.ai's overall scoring system. CyberGym is currently displayed for reference but excluded from the scoring formula, so it does not directly affect overall rankings.

About CyberGym

Year

2026

Tasks

1,507 vulnerability analysis instances

Format

Vulnerability reproduction and PoC generation

Difficulty

Real-world cybersecurity

CyberGym includes 1,507 benchmark instances from historical vulnerabilities across 188 large software projects. BenchLM stores CyberGym as a display-only agentic security benchmark when exact provider comparison values are published.

CyberGym: Evaluating AI Agents' Real-World Cybersecurity Capabilities at Scale

BenchLM freshness & provenance

Version

CyberGym 2026

Refresh cadence

Quarterly

Staleness state

Current

Question availability

Public benchmark set

CurrentDisplay only

BenchLM uses freshness metadata to decide whether a benchmark should still be treated as a strong differentiator, a benchmark to watch, or a display-only reference. For the full scoring policy, see the BenchLM methodology page.

FAQ

What does CyberGym measure?

A cybersecurity task benchmark for evaluating defensive cyber workflows and vulnerability-oriented agent performance.

Which model scores highest on CyberGym?

Fugu Cyber by Sakana AI currently leads with a score of 86.9% on CyberGym.

How many models are evaluated on CyberGym?

20 AI models have been evaluated on CyberGym on BenchLM.

Compare Top Models on CyberGym

Fugu Cyber vs GPT-5.6 SolGPT-5.6 Sol vs GLM-5.3GLM-5.3 vs Claude Mythos 5Claude Mythos 5 vs DeepSeek V4 Pro 0813
Last updated: August 22, 2026 · BenchLM version CyberGym 2026

Know when it’s worth switching models

The model to choose, the cheaper alternative, and the release we would wait on.

Read a sample issue

Join 2,000+ readers.

One email each week. Unsubscribe anytime.