Xbow: AI agents ran a six-day exploit campaign on gpt-5.6-sol, costing ~$40,000 in API fees
rez0__ · x · 2026-10-14
Security firm Xbow reports its AI agents, powered by OpenAI's gpt-5.6-sol, autonomously searched source code, tested hypotheses, built proofs of concept, and developed exploit stages over a six-day campaign. Reconstructing usage from archived Codex logs puts the total at roughly $40,000 in API charges. rez0 adds that 5.6-sol remains "goated" for this kind of work — a concrete data point on both the offensive security capability of frontier models and what such campaigns now cost.
More from Models
- Gemini 4 praised as a big leap — but its generated UI is hilariously over the top — BraydonDymm · 2026-10-14
- Gemini 4 takes the crown on GamowLabs' RareBench and LabBench, dethroning Opus 5.5 by a healthy margin — sahandsharif · 2026-10-14
- Muse Wonder's under-discussed edge: first frontier lab to report 'big model smell' anti-benchmaxxing metric — alexandr_wang · 2026-10-14
- TV noise crashes voice agents: task completion falls from 38.7% to 8.6% across 14 stacks — solyarisoftware · 2026-10-14
- GPT-5.6 Sol finds critical Chrome vulnerability after 6 days and $40,000 of compute — gdb · 2026-10-14
- Early Muse Wonder impressions: no personality regressions, subtle bug catches and better UI sense — alexandr_wang · 2026-10-14