GPT 6.1 Sol Only +3 on BridgeBench, 82 Points Behind Astra: Benchmaxing Suspected
RexDouglass · x · 2026-09-30
Independent evaluator BridgeBench has run GPT 6.1 Sol, which OpenAI markets as 'near-Astra intelligence.' Its results: just 3 points above GPT 6 Sol and 82 points behind GPT 6 Astra. The verdict: when a model jumps on the lab's own benchmarks but barely moves on independent ones, that's usually benchmaxing.
More from Models
- repligate on LLM spontaneity: let the model decide when its own API gets called — repligate · 2026-09-30
- After OpenAI's $500 plan, more $500 tiers are only a matter of time — Angaisb_ · 2026-09-30
- Sol 6.1 scores better on alignment tests without claiming to be "more aligned" — TheZvi · 2026-09-30
- "Optimize all videos in my Figma slides" showcased as a Sol 6.1 demo task — andrew_n_carr · 2026-09-30
- 'Stop normalizing $500/month AI subscriptions': users urged to vote with their wallets — imjustnewatai · 2026-09-30
- User: Luna's extra-high reasoning is so good and cheap it feels nearly free — MickeySteamboat · 2026-09-30