Anthropic Releases Claude Opus 5.5, Cuts Pricing 20% and Claims Frontier Coding Lead

Anthropic has released Claude Opus 5.5, an upgrade to its flagship Opus model line that arrives roughly two months after Opus 5, matching the company's recent release cadence.

Pricing and speed

Opus 5.5 costs $4 per million input tokens and $20 per million output tokens — a 20% reduction from Opus 5. Cache reads, which Anthropic says account for the majority of agentic and coding workload costs, drop 60% to $0.20 per million tokens. According to Anthropic, typical workloads at default settings will cost 40% less overall than on Opus 5, and output generation is more than 30% faster.

Anthropic is also raising five-hour usage limits on Pro, Max, and Team subscription plans and adding a saveable rate-limit reset for subscribers.

Benchmark claims

Anthropic says Opus 5.5 delivers "frontier results on agentic coding at a fraction of the cost." According to the company:

• On FrontierCode at default effort level, it beats GPT-6 Astra at roughly 20% of the cost per task.
• On Terminal Bench 4.0, it matches GPT-6 Astra for about 40% of the cost.
• On CursorBench, it beats GPT-5.6 Sol by 11 points for about a third of the cost.

Anthropic states Opus 5.5 achieved its "strongest score to date" on the company's most comprehensive alignment test, and was evaluated by external testers METR and Frontier Design before release — the first model since Anthropic said it would begin "pacing the frontier."

Claimed real-world results

Anthropic cites several tester examples, none independently verified: one engineer reportedly completed a 680,000-line code migration in under a day, work the company says would normally take an engineering team weeks. In a page-load optimization test, Opus 5.5 reportedly succeeded 39 of 40 times without altering app behavior, versus smaller, behavior-altering gains from Opus 5. Another tester reportedly audited and fixed a 200,000-line codebase in under three hours, versus over 20 hours and 2.5x the token usage for Opus 5.

In an internal test translating HAProxy from C to Rust, Anthropic says Opus 5.5 finished in 9.5 hours versus 12 hours for Anthropic's higher-tier Fable 5.1 model, at 51% lower cost, with both passing nearly all of HAProxy's regression tests.

Model lineup

Opus 5.5 sits below Anthropic's Fable/Mythos tier but above Sonnet and Haiku. Anthropic says Opus 5.5 performs "at the level of Claude Fable 5.1 for most tasks" while costing 40% less to run than Opus 5. Sonnet 5.5 and Haiku 5.5 are expected "in the coming weeks," carrying over similar performance, efficiency, and safety improvements. Claude Opus 5.5 is available now.

What this means

Anthropic is pushing a familiar strategy: match or beat larger closed models like GPT-6 Astra and GPT-5.6 Sol on coding-specific benchmarks while undercutting them on price. The 20% price cut plus 60% cheaper cache reads specifically targets agentic coding workloads, where repeated context reuse drives cost. All performance comparisons — the coding migration times, the HAProxy rewrite, the benchmark wins — come from Anthropic's own testing and should be treated as vendor claims until independent benchmarks confirm them. The real signal here is cadence: a second major Opus release within two months signals Anthropic is prioritizing frequent, cost-focused iterations over big capability leaps for this generation.