Don't trust evals: early adopter says Astra is the first model to cross his trust threshold
sandersted · x · 2026-09-04
A developer shares first-hand impressions of gpt-6 Astra: "don't trust evals" — benchmarks are often sloppy and miss what matters, and small helpful behaviors like asking clarifying questions can tank a score. Still, he argues Astra is the real deal: imperfect, but the first model to cross his personal trust threshold.
Related event: OpenAI launches flagship GPT-6 Astra to rave reviews and pricing debates(190 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