AI-generated editorial illustration; not a documentary photograph.
Anthropic introduced Claude Opus 5.5 on Tuesday as the first model in its Claude 5.5 family, positioning it for broader software-development work than code completion alone. The company says the model can assist across design specification, code generation, debugging and testing, with operating costs around 40% below Opus 5.
Built for longer tasks
Anthropic claims Claude Opus 5.5 is particularly suited to codebase-wide migrations and audits. An early tester reported that it audited and fixed a 200,000-line codebase in under three hours, compared with more than 20 hours for Opus 5, while Opus 5 used 2.5 times as many tokens. Those results are reported claims rather than independent tests.
Reported engineering results
In an Anthropic internal comparison, Claude Opus 5.5 translated HAProxy from C to Rust in 9.5 hours, versus 12 hours for Claude Fable 5.1. Both versions reportedly passed nearly all of HAProxy’s regression tests, while the Opus 5.5 run cost 51% less. Anthropic also says the model produced output more than 30% faster than Opus 5 and required fewer tokens for higher-quality work.
Pricing and safeguards
Anthropic lists Claude Opus 5.5 at $4 per million input tokens and $20 per million output tokens, with cache-read prices reduced by 60% for token-billed usage. The company says Frontier Design and METR tested the model before release and that external evaluation and alignment testing informed its safeguards.
Human verification remains necessary
The model’s broader ability to run commands, inspect failures, change files and verify results also increases the consequences of mistakes. The New Stack’s reporting highlights the risk that a task can appear complete without being correct, and notes that some requests are routed to other models when safeguards activate. Cybersecurity requests are generally redirected to Opus 4.8, while biology and frontier-LLM-development requests can be redirected to Opus 5; approved organizations can apply for Anthropic’s Life Sciences Verification Program.
Source: Claude Opus 5.5 wants to finish your coding tasks, not just start them
SIGNAL TITANS
Leave a Reply