Claude Fable 5.2 vs. GPT-6 Astra: A Lower List Price Didn't Mean a Lower Bill
The New Stack, Saturday, November 14th, 2026
In coding tests, Claude Fable 5.2 matched or beat GPT-6 Astra but cost far more because it used many more tokens.
Jessica Wachtel ran Anthropic's Claude Fable 5.2 and OpenAI's GPT-6 Astra once each on three coding tasks: a bug fix, a code review and code writing.
Both models scored 50 of 50 on the bug fix and code writing, while Fable 5.2 led the code review 49 to 45 by proving more planted defects.
Although Fable 5.2 has lower per-token list prices, it used about five times as many tokens, and on the two tied tasks it cost 2.4 times as much as Astra and took 1.9 times as long.
The author would start with Astra for similar tasks and consider Fable 5.2 for careful code reviews.