Claude Opus 5.5: price, context window and specs

What Anthropic says it's for

Anthropic's models overview describes Opus 5.5 as the model "for long-running agentic coding and knowledge work" and tells developers to start with it for most workloads. In its announcement, Anthropic says Opus 5.5 generates output more than 30% faster than Opus 5 and costs 40% less than Opus 5 on typical workloads at default settings, because it's cheaper per token and uses fewer tokens per task. Those are Anthropic's own figures.

Pricing details

• Base: $4 per million input tokens and $20 per million output tokens, 20% below Opus 5's $5 / $25.
• Prompt caching: 5-minute cache writes $5, 1-hour writes $8, cache reads $0.20 per million tokens (5% of the input price).
• Batch API: 50% off input and output ($2 / $10).
• The full 1M-token context is billed at the standard rate, with no long-context surcharge.
• Fast mode (research preview): $8 input / $40 output per million tokens. On the API side it's Claude API only; Anthropic says it's also in Claude Code.

Changes that can break existing API code

Per Anthropic's model page, adaptive thinking is always on: requests with thinking disabled return a 400 error, so you control depth with the effort parameter instead. Forced tool use (tool_choice any or tool) also returns an error, thinking blocks are tied to the model and conversation that produced them, and the older computer_20251124 tool isn't accepted on the Claude API or Google Cloud. Opus 5.5 defaults to medium effort, one level lower than Opus 5, so a request that omits effort now runs at medium.

Using Opus 5.5 in Claude Code

• Claude Code v2.1.280 (22 Sep 2026) added Opus 5.5 as the default Opus model and switched Pro and Team Standard from Sonnet to Opus as the default.
• The opus alias resolves to Opus 5.5 on the Anthropic API, Claude Platform on AWS, Amazon Bedrock and Google Cloud. On Microsoft Foundry it resolves to Opus 4.6, per the model configuration docs.
• Switch with /model opus or start with claude --model opus. The opusplan setting uses Opus in plan mode and Sonnet for execution.
• Claude Code starts Opus 5.5 at medium effort. Change it with /effort (low, medium, high, xhigh or max).
• On launch day ClaudeDevs said five-hour session limits in Claude Code rose 20% and that Opus 5.5 goes 25% further within limits because it's priced lower.

Independent benchmarks

Artificial Analysis lists Claude Opus 5.5 (max effort, default fallback) at 58 on its Intelligence Index v4.3.2, at an average of $5.98 per Intelligence Index task, as of 13 October 2026. In its 12 October post, it reported Claude Code with Opus 5.5 at 66 on its Coding Agents Index, behind Claude Code with Sonnet 5.5 at 68. Anthropic's own table reports 66.4% on Terminal-Bench 4.0 at xhigh effort. Scores change as labs re-run evaluations.