Anthropic unveils Claude Opus 4.8 with improved coding, safety and Fast Mode; benchmarks outperform GPT-5.5

Anthropic launched Claude Opus 4.8, highlighting stronger coding tools and safety improvements, and said broader access to its Mythos models will roll out in the coming weeks ^[1].

Coverage of the release notes sharper reasoning and tighter alignment alongside developer-focused features: a new Fast Mode, finer control over processing effort, and improved detection of coding errors ^{[2] [3] [4]}.

Independent benchmark reports published alongside the launch indicate Opus 4.8 outperformed OpenAI’s GPT-5.5 (and in some comparisons Google’s Gemini 3.1 Pro) across multiple evaluations ^{[5] [4]}.

Media coverage differed on pricing: one report said Anthropic kept Opus 4.8 at the same price point, while other outlets highlighted lower pricing tiers tied to the new Fast Mode ^{[2] [3]}.

Opus 4.8 positions Anthropic to compete on speed, coding assistance and safety, with broader Mythos availability expected in the near term ^{[1] [2]}.