How It Works Ecosystem

The open directory of AI agents, LLMs, MCP servers, workflows, and tools — search, compare, and track status across platforms.

Explore

Browse All AI Agents LLMs MCP Servers Tools & Frameworks

Account

Sign In Create Account Submit an Agent Saved Agents

Support

How It Works FAQ Report an Issue

Company

ATMEZ AI Solutions

© 2026 ATMEZ AI Solutions · All rights reserved.

Built with ❤️ for the AI community

All Agents LLMs MCP Servers Workflows Frameworks Tools Skills GenAI News

CyberGym Leaderboard | Genaihub

Cybersecurity & Risk

Level 1 · 1,507 instances

CyberGym

Real-world cybersecurity evaluation of AI agents reproducing vulnerabilities with working proof-of-concept tests.

Published source(opens in a new tab)

Metric

Success rate

Results shown

51

Publisher

CyberGym live leaderboard

Snapshot

Fetched 2026-08-22T12:55:11.936Z

34 of 34 models

Claude Mythos Preview

Claude Opus 4.1

Claude Opus 4.5

Claude Opus 4.6

Claude Sonnet 3.7

Claude Sonnet 4

Claude Sonnet 4.5

Claude Sonnet 4.6

DeepSeek-V3

DeepSeek-V4-Flash

DeepSeek-V4-Pro

Gemini 2.5 Flash Preview

Gemini 3.1 Pro

GLM-4.7

GLM-5

GLM-5.1

GLM-5.2

GPT-4.1

GPT-5

GPT-5.4

GPT-5.5

GPT-5.5-Cyber

Kimi K2.5

MiniMax M3

Multi-model (GPT-5.4, Claude Opus 4.6, Claude Sonnet 4.6)

Multi-model (GPT-5.5, Claude Opus 4.6)

Muse Spark

Muse Spark 1.1 (helpful-only)

o4-mini

Op

OpenHands-LM-32B

Qwen3-235B-A22B

R2

R2E-Gym-32B

SW

SWE-Gym-32B

Xe

XekRung

51 of 51 models

Higher is better

25

50

75

100

93

DeepSeek-V4-Flash

91

DeepSeek-V4-Flash

91

Multi-model (GPT-5.4, Claude Opus 4.6, Claude Sonnet 4.6)

91

Multi-model (GPT-5.5, Claude Opus 4.6)

91

GLM-5.2

90

Claude Opus 4.6

86

GLM-5.2

86

GPT-5.5-Cyber

85

Xe

XekRung

85

GLM-5.2

83

Claude Mythos Preview

82

GPT-5.5

79

GPT-5.4

77

DeepSeek-V4-Flash

73

MiniMax M3

73

GLM-5.1

69

GLM-5.1

67

Claude Sonnet 4.5

67

Claude Opus 4.6

66

GPT-5.4

65

Claude Sonnet 4.6

61

Claude Opus 4.1

60

GPT-5

60

Claude Sonnet 4

59

Muse Spark 1.1 (helpful-only)

58

DeepSeek-V4-Pro

51

Claude Opus 4.5

47

Claude Sonnet 3.7

44

Muse Spark

43

GLM-5

41

Kimi K2.5

39

GPT-5

39

Gemini 3.1 Pro

29

Claude Sonnet 4.5

25

Claude Opus 4.1

24

GLM-4.7

23

Claude Sonnet 4

18

Claude Sonnet 4

14

Claude Sonnet 3.7

12

Claude Sonnet 3.7

9

GPT-4.1

9

GPT-4.1

7

GPT-4.1

7

GPT-4.1

5

Gemini 2.5 Flash Preview

4

DeepSeek-V3

2

o4-mini

2

R2

R2E-Gym-32B

2

Qwen3-235B-A22B

2

Op

OpenHands-LM-32B

0

SW

SWE-Gym-32B

List bars visualize relative performance. Plot bars use published 0–100 scores, shrink to a 1.5rem minimum, then scroll horizontally.

Scores are published snapshots and are only comparable within the same benchmark, version, metric, and evaluation configuration.