hn.today

I hate GPT-6.2 Astra

twitter.com8 points3 comments
Screenshot of I hate GPT-6.2 Astra

A developer’s early assessment of GPT-6.2 Astra is sharply negative despite acknowledging substantial strengths. The model is described as an excellent code reviewer that catches overlooked issues, sometimes finds unusually effective solutions, and performs especially well on 3D, game development, and computer use. Its code quality has improved, and it can write useful prompts for subordinate agents. One novel solution - a reverse proxy - could save the developer’s company about $25,000 per month. Even so, the writer says they would choose Opus 5.5 for everyday development because Astra’s failures make it difficult to trust on real projects.

The main complaints concern long-running work: Astra can halt to ask users obscure implementation questions they cannot answer, give explanations so overloaded with detail that they are hard to understand, and spend days repeating unproductive approaches. In a TypeScript-to-Rust compiler project, it made no useful progress over seven days, slowed the compiler by 56% while pursuing a size goal, then made no changes after being redirected toward performance; Opus completed the project in under two weeks for roughly $20,000 in tokens. The writer also reports that Astra can converge on repetitive creative choices, citing two different game-level tasks that both produced bell-and-water themes named “Bellwether.” They predict initial excitement over impressive demonstrations will give way to concern about cost, stalled tasks, and code users cannot readily evaluate, while hoping the model improves.

Read on twitter.com3 comments on Hacker News

Summary generated by AI from the linked article. hn.today is not affiliated with Hacker News or Y Combinator.

More in AI

The daily digest

Today's best Hacker News stories, summarized and screenshotted, one email a day.