AivexaNewsSearch
AI news for builders and product teamsChecked every hour

[AINews] Claude Opus 5.5, the new default model for AINews — and everybody cuts prices 40-50%

Collected Sep 30, 2026

Anthropic shipped Claude Opus 5.5, described as the first model in a new Claude 5.5 family. OpenAI released GPT-6 Sol and Luna about an hour later.

Anthropic said Opus 5.5 "performs at the level of Claude Fable 5.1 for most tasks, and costs 40% less to run than Opus 5." According to the company's posts, it leads on agentic coding, computer use, and knowledge work, is about 30% faster and about 40% cheaper per task than Opus 5, and puts the most important information up front while following user writing rules. Opus 5.5 is now the default in Claude Code and the Claude app, including Cowork, with default effort set to medium. It is live in Claude Code, the Claude Platform API, and Claude Tag for Slack. Anthropic said Sonnet 5.5 and Haiku 5.5 follow "in the coming weeks," contradicting rumors that Haiku was discontinued. It is the first Opus with Fable 5.1-class safeguards on cyber, bio, and frontier LLM development; flagged requests fall back to another model, and Anthropic said it is "working to reduce incorrect flags." Subscription changes include 5-hour session limits up 20%, with lower pricing meaning limits go 25% further; Pro, Max, and Team users get a banked rate-limit reset.

On pricing, ValsAI reported list token pricing was cut 20%, from $5/$25 to $4/$20 per 1M input/output tokens. Vals noted Opus 5.5 often uses more tokens, especially on coding, partly offsetting the lower sticker price. Artificial Analysis said that at max effort Opus 5.5 costs $5.98 per Intelligence Index task versus $5.86 for Opus 5 (max); it attributed the difference to higher token usage and the base-price cut. Cline reported Opus 5.5 beats Fable 5.1 on the Artificial Analysis Intelligence Index at about 2.5x lower cost. Prompt caching is not broken when switching effort mid-session on Claude Code v2.1.280+. A claim by @theo that Opus 5.5 is smaller than Opus 5 was not confirmed in official posts. A system card was published at launch. The model was spotted in Claude Code shortly before the announcement.

Read at Latent Space

Based on reporting from the original publisher. Visit the source for full context and later updates.

Publisher excerpt

overshadowing more efficient GPT6 models from OpenAI