VECTOR WIREAI INTELLIGENCE
NVDA$1,847+3.2%MSFT$512+1.1%GOOGL$199-0.4%META$728+2.7%AMD$184-1.2%TSM$212+0.6%PLTR$98+4.1%AI IDX4,821+1.9%
PKT
SEEDRefresh Models Deals Regulatory Sources

cybergym

22 models tracked

CyberGym — Cybersecurity vulnerability-reproduction agentic benchmark

#ModelVendorBest scoreRunsLast seen
1GLM-5.3Z.ai84.522026-08-23
2Claude Mythos 5Anthropic83.822026-08-23
3GPT-5.6 SolOpenAI83.612026-08-17
4Gemini 3.5 Flash CyberGoogle83.212026-08-23
5Claude Opus 4.8Anthropic83.142026-08-23
6GPT-5.5OpenAI81.822026-08-23
7Kimi K3Moonshot8012026-08-13
8DeepSeek-V4-FlashDeepSeek76.762026-08-23
9Claude Opus 4.6Anthropic73.842026-08-23
10Claude Opus 4.7Anthropic73.822026-08-23
11GLM-5.1Z.ai68.732026-08-23
12GPT-5.4OpenAI66.322026-05-18
13Muse Spark 1.1Meta5912026-07-13
14DeepSeek-V4-ProDeepSeek52.722026-08-13
15Claude Opus 4.5Anthropic50.632026-07-29
16GLM-5Z.ai48.332026-05-18
17Muse SparkMeta43.512026-07-16
18Kimi K2.5Moonshot41.352026-08-23
19Gemini 3 ProGoogle39.922026-08-24
20Gemini 3.1 ProGoogle38.822026-05-18
21GLM-4.7Z.ai23.512026-06-13
22DeepSeek-V3.2DeepSeek17.342026-06-15

Best tracked score per model (default configuration; source-attributed and verification-tiered). Open a model for its full benchmark surface, provenance and pricing.