Anthropic launches Claude Opus 5.5, cutting price 40% versus Opus 5

2026-09-22

Anthropic released Claude Opus 5.5 (claude-opus-5-5) for long-running agentic coding and knowledge work, with a 1M-token default context window, 128k max output tokens, and always-on adaptive thinking (thinking cannot be disabled via the API; effort controls thinking depth instead). Pricing is $4/$20 per MTok versus Opus 5's $5/$25. Anthropic's own announcement states Opus 5.5 "performs at the level of Claude Fable 5.1 on most work and costs 40% less to run than Opus 5." Fast mode is available as a research preview on the Claude API, and computer-use on this model now requires a new toolset (computer_toolset_20260801) on the API and Google Cloud.

Significance 4: A new flagship Anthropic model with a large price cut (40% vs Opus 5) and a lab-reported claim of Fable-5.1-level quality plausibly reshapes the daily-driver and coding-agent verdicts, which currently run on the now two-versions-old Opus 4.8; the quality claim caps significance below 5 until independently reproduced.

technology financial business Release Pricing change Tooling / agents

Implications · machine-drafted, not owner judgment

A materially cheaper Opus tier at claimed Fable-5.1-level quality raises pricing pressure on GPT-5.2 and Gemini 3 Pro and could shift the daily-driver and coding-agent picks once independent benchmarks land, since both current verdicts run on the older, pricier Opus 4.8. Always-on thinking (no disable option) also changes cost-per-call math for high-volume agentic use, since callers can no longer opt out of the reasoning tax. For solo AI-engineering work like this tracker, a cheaper frontier-quality coding model modestly lowers the cost floor for hands-on build work.

Watch for

Independent SWE-bench Verified / Terminal-Bench scores for Opus 5.5 within a month
Whether Artificial Analysis or LMArena place Opus 5.5 at or above Fable 5.1
Whether the daily-driver or coding-agent verdict changes

Sources

primary Claude Platform release notes retrieved 2026-09-24
primary Introducing Claude Opus 5.5 retrieved 2026-09-24

Models

Claude Opus 5

Affected axes

Cost · Coding · Agentic / tool use · Hard reasoning · Output speed

← Back to feed