3 Weeks with GPT-6.2 Astra Internal Checkpoint: Less Lazy, Less Coding Slop
SIGKITTEN · x · 2026-11-18
Developer daidaijiayu reports testing GPT-6.2 Astra from an internal checkpoint for about three weeks before his vacation. Compared to 6 Astra, the improvements:
- Less lazy
- Writes less "coding slop"
- Generally easier to work with
Still needs work:
- Writing lacks high-level narrative progression
- Implicit instruction following (understanding what's not verbatim in the prompt) remains imperfect
He notes this impression differs sharply from today's sentiment on X. The post was retweeted by SIGKITTEN (18.5k followers). Note: 6.2 Astra is unreleased; these are early impressions of an internal build.
More from coding & agent
- Dev builds Kasane Valley with Blender + three.js, showcasing 'GPT 6.2 Astra Ultra' workflow — OpenAIDevs · 2026-11-18
- User files a Chase complaint addressed to the bank's AI agent, with a full image transcript — AaronBergman18 · 2026-11-18
- Anthropic on agent memory: scaling 'dreaming' to 128 sessions lifts score from 13% to 48% — dl_weekly · 2026-11-18
- 10 people wrote 15,000 e2e tests in 2 weeks using this automation system — jdluk87 · 2026-11-18
- Dev ships local-first "central nervous system" for AI with knowledge graph and rollback — bankhosa_01 · 2026-11-18
- Derive MCP brings 31+ trading tools for options, perps and spot to AI assistants — modelcontextprotocol · 2026-11-18