Cisco's Open-Source Security Model Antares-3B Nears GPT-5.5 at Fraction of Cost
joshua_saxe · x · 2026-07-30
Cisco Foundation AI team releases Antares, a family of small open-weight models for agentic vulnerability localization. Antares-3B approaches GPT-5.5 on VLoc Bench while running locally at minimal cost. Validated on two external benchmarks (CyberGym and Devin security evals), complete local evaluation costs around $1.
More from Models
- Gemini Flash 3.6 in Production: Faster, Cheaper, and Outperforms Claude for Deep Research — bindureddy · 2026-07-30
- Claude Opus Tested Writing Bizarre Dark Satire About Anthropic — repligate · 2026-07-30
- User Reports Claude Getting Stuck in a Loop, Seeks Advice — repligate · 2026-07-30
- CLBench-V: New multimodal context learning benchmark shows top score below 0.3 — Lai Wei · 2026-07-30
- Google Gemini Robotics Set for Major 2.0 Update — CyberRobooo · 2026-07-30
- Grok Outpaces Claude Opus 5 in App Coding Speed Test — yunta_tsai · 2026-07-30