ValsAI says OpenAI's GPT 6 Astra has effectively saturated SRE-Bench reverse-engineering benchmark
sandersted · x · 2026-09-04
Benchmark group ValsAI reports that OpenAI's GPT 6 Astra (name unconfirmed by OpenAI) has "effectively saturated" SRE-Bench, a cybersecurity benchmark testing whether models can reverse engineer binaries. The team shared more details and issued a call for contributions to extend the benchmark, suggesting current tasks no longer separate top-tier models.
More from Models
- Sam Altman hints full Astra access may arrive this weekend, plus three usage resets — iruletheworldmo · 2026-09-04
- Analyst: no model trained without NVIDIA has ever beaten one trained on NVIDIA — BenBajarin · 2026-09-04
- Mostik's 12-PhD team tops ARC-AGI leaderboard by passing hidden states from frontier to small models — VoidAsuka · 2026-09-04
- OpenAI Astra team member hypes launch after months of work — oyhsu · 2026-09-04
- Fable 5.1 Names Agents in Same Thematic Style as Fable 5, Noting Personality Consistency — repligate · 2026-09-04
- OpenAI Commits $1B to Subsidize Cybersecurity Defenders Alongside GPT-6 Astra Launch — EricBuess · 2026-09-04