Create Your Personal Python & AI Coding Learning Plan →
OpenAI just released GPT-6 Astra, so Dan Bader, Martin Breuss, and Philipp Acsany, jumped on a call to compare notes after a weekend with the new model. No benchmarks, no demos, just three Python developers talking about how Astra feels in day-to-day work: how it stacks up against Claude Fable 5.1, why the low effort setting often gives better results, what the new "don't prompt engineer" guidance means for you, and whether the excitement will survive the honeymoon phase.
Download your free Python Cheat Sheet here:
Free Python Skill Test with instant level + learning plan:
Want to learn faster? Become a Python Expert with unlimited access to 5,000+ tutorials, videos, and exercises:
Chapters:
00:00 A new model from OpenAI
00:33 Dan's first impressions: Codex, Rust, and low effort
03:59 Martin's first impressions: text, voice, and collaboration
06:30 OpenAI's marketing vs the real impact
08:23 Has prompting changed?
09:48 Stop prompt engineering: new guidance from OpenAI and Anthropic
12:10 What "aligned" feels like in practice
13:59 Where Astra still trips up
16:02 Astra vs Fable 5.1
17:11 Claude-isms and a refreshingly terse model
18:34 Why low effort beats high effort
20:38 Recurrent transformers and hidden thinking
23:05 Rewriting a Python codebase in Rust
24:46 The honeymoon phase
25:51 Developer relations: OpenAI vs Anthropic
28:13 Why model releases feel exciting again
29:32 Should you try GPT-6 Astra?
33:19
|
2026/09/08
youtube
