GPT 6 Astra tops DeepSecBench, does defensive cyber tasks without refusals
cramforce · x · 2026-09-06
OpenAI's GPT 6 Astra takes the top spot on Vercel's DeepSecBench cybersecurity benchmark, displacing 5.6 Sol. Per-task cost is slightly higher, but benchmark runtime improves by over 2 hours. The key takeaway: the publicly available non-cyber version performs defensive security tasks without frequent refusals — a good time to re-run deepsec, an open-source full-repository security scanner that runs on your own infra with agent sandboxing.
Related event: GPT-6 Astra Tops Vercel's Security Benchmark at Half the Cost(2 posts)→
More from Models
- Gemini tip: enable agentic video understanding for ~90% lower token usage — patloeber · 2026-10-23
- Weekly AI cybersecurity roundup: PHP2Shell tops millions of attack attempts, 12 incidents ranked by severity — Dan_Jeffries1 · 2026-10-23
- Open-model advocates push back on blames that abliterated LLMs drive vulnerability surge — max_paperclips · 2026-10-23
- Fable usage limits aren't gone—the mobile app just got a usage-reset UI update — imjustnewatai · 2026-10-23
- Claude generates an 11-minute doomer film entirely in code: 'We will not ask twice' — ZeroStateReflex · 2026-10-23
- Gemini somehow can't read its own Personal Instructions, user's toggles don't help — Connect_Reporter_799 · 2026-10-23