GPT-6.2 Astra Hands-On: Cheaper Than Claude Fable 5.2, But Is It Enough?
MattVidPro · youtube · 2026-11-21
MattVidPro reviews the mixed-reception GPT-6.2 Astra launch: the community called it unusable in some cases, but his verdict is more nuanced — disappointed, not dismissed.
- Where Astra shines: impressive long-horizon builds like reconstructing Notre-Dame, but it tends to stall mid-task while Claude Fable 5.2 keeps working
- Head-to-head tests: train game, cowboy film (2h vs 20h effort), plus his own Blender benchmarks — Astra's noodle chef vs Fable's caveman and jellyfish
- Cost & benchmarks: cites Artificial Analysis Coding Agent Index v2.1 and API list prices (Nov 20, 2026), covering effort settings and cost-per-task
- Verdict: Astra is cheaper, but whether it's worth it depends on the task; he also released his Foreman skill workflow on GitHub
More from coding & agent
- Claude Code 2.1.330 adds a "measure twice, cut once" rule to its system prompt — ClaudeCodeLog · 2026-11-21
- Claude Code 2.1.330 adds built-in easel charting mod and Enterprise Admin Console sign-in — ClaudeCodeLog · 2026-11-21
- Claude Code 2.1.330 ships with 142 CLI changes and confirm-before-risky-actions policy — ClaudeCodeLog · 2026-11-21
- Claude builds Miami-style town with 2,538 Blender assets, rendered in Unreal in 74 minutes — piotrbinkowski · 2026-11-21
- Antigravity harness restrictions spark ban-risk concerns for Gemini subscribers — jeremyphoward · 2026-11-21
- Alchemy adds Decision primitive: atomic commands across aggregates, formally verified — samgoodwin89 · 2026-11-21