Every's Vibe Check: GPT-6 Astra Drafts Its Own 3,000-Word Review, Still Trails Fable on Product Instincts
danshipper · x · 2026-09-04
Every published its detailed "Vibe Check" of GPT-6 Astra vs Anthropic's Fable 5.1.
- Same-night turnaround: the team learned of Astra's launch at 3 am; by 7 am a 3,000-word draft review — written by Astra from a single prompt — was ready, complete with sections, headline and screenshots. CEO Dan Shipper mistook it for human-written; staff writer Katie Parrott quipped "RIP my job."
- Verdict: Astra is "a big upgrade with some bad habits," impressing on writing, operating software and visual design, while Fable still has better instincts for building a product.
- The full report covers every benchmark plus a final verdict, with a live head-to-head camp for paid members.
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