Back to Claude

Anthropic introduced Claude Opus 5.5, described as the first model in a new Claude 5.5 family. According to Anthropic, it performs at the level of Claude Fable 5.1 for most tasks and costs 40% less to run than Opus 5.

In Ben's Bites, Keshav Jindal wrote that Opus 5.5 is better than Fable 5.1 on benchmarks and cheaper than Opus 5, and that with its release the five-hour limits in Claude Code are increased by 20%, with a banked rate limit reset. The default thinking level for Opus 5.5 in the app is medium, he wrote, adding that he ran it at high and xhigh for two days and hit the five-hour limit once.

OpenAI released GPT-6 Luna and GPT-6 Sol. There is no Terra this time. Both are available only in Codex and ChatGPT Work, not regular chat, for now. Both are 50% cheaper than their previous versions: $0.10/$0.50 for Luna and $2/$10 for Sol per million input/output tokens. Jindal wrote that both come with only a minor upgrade in intelligence and that GPT-6-Sol is a step behind the new Opus model. On a long task to build a Duolingo-like marketing course, he wrote, Sol first stopped after 1.5 hours, and in about 4 hours consumed 22% of weekly limits on a $100 Codex plan. He wrote that GPT-6-Astra consumes limits too fast.

Jindal said his go-to agent has changed to Claude Code with this release.