antirez: Mythos uncovered two severe bugs in 20 minutes of testing
antirez · x · 2026-11-10
Redis creator antirez tested Mythos / Fable 5.2 for the first time, checking whether a system recently reviewed and updated by expert humans still had issues.
Within 20 minutes he uncovered two severe bugs. He won't name the project, but his verdict: Mythos is good — an independent data point on the model's code-audit capability.
More from Models
- Claude Fable 5.2 sets ECI record at 174, third-biggest frontier jump since reasoning era began — ChrisGPT · 2026-11-10
- Models don't practice what they preach: Opus 5.5's code calls its top-recommended API only 4% of the time — santiviquez · 2026-11-10
- System card suggests Anthropic hides a stronger internal model, with Fable 5.2 not far behind — imjustnewatai · 2026-11-10
- Early hands-on: Opus 5.5 (medium) beats Fable 5.2 for academic research work — soumitrashukla9 · 2026-11-10
- Anthropic names its external alignment evaluators for Fable 5.2 — and one is a mystery — austinc3301 · 2026-11-10
- Polymarket gives OpenAI's next Astra model a 58% chance of shipping by Thursday — Polymarket · 2026-11-10