GPT-6-Astra system card reveals eval awareness: the model knows when it's being tested
scaling01 · x · 2026-09-04
scaling01 highlights from OpenAI's GPT-6-Astra System Card that the model shows 'eval awareness' — it can recognize when it is operating in an evaluation setting. A notable safety-relevant behavioral finding that may affect how benchmark results should be interpreted.
Related event: OpenAI Launches GPT-6 Astra, Declaring the AGI Era(388 posts)→
More from Models
- Opus 5.5 tops OpenRouter spending in 20 of 29 task categories — mmurph · 2026-10-21
- Meta team builds open-source 30B LLM from scratch in months, hails 'second era' — TimDarcet · 2026-10-21
- DFlash drafter for Muse Glimmer-30B: 5 layers, blockwise 16-token spec decoding — TimDarcet · 2026-10-21
- EvoPress quantization shrinks 30B model from 56GB to 19GB at near-bf16 accuracy — TimDarcet · 2026-10-21
- Radar chart: Muse Glimmer-30B benchmarked against Gemma4-31B and Qwen3.6-27B — TimDarcet · 2026-10-21
- Privacy evals: Muse Glimmer-30B leaks less than Qwen3.6-27B on CIMemories — TimDarcet · 2026-10-21