Anthropic releases Claude Opus 5.5 at 40% lower cost with stricter safeguards

According to Anthropic, Claude Opus 5.5 achieves performance equivalent to Fable 5.1 at 40% lower cost and includes stricter safeguards against sandbox escapes and risky behaviors. The release received 778 points on Hacker News, though multiple users reported the model performed worse than expected and made more errors than prior versions despite unchanged usage patterns.

Update (2026-09-24): According to CIO, OpenAI released GPT-6 Sol and Luna with per-token costs half those of GPT-5.6, while Anthropic released Claude Opus 5.5 delivering Fable 5.1 performance at 40 percent lower cost. OpenAI attributed cost reductions to improvements in ...