Anthropic Launches Claude Opus 4.8 With Gains in Coding and Honesty
Source: MacRumors
Anthropic today announced the launch of its latest AI model, Claude Opus 4.8. Anthropic describes the model as a “more effective collaborator,” with improvements in agentic coding, multidisciplinary reasoning, agentic computer use, knowledge work, and agentic financial analysis.
Reliability, Honesty, and Alignment
Testers have found Opus 4.8 to be “more reliable and sharper in its judgement” when performing agentic tasks, and the model also shows gains in honesty. Early testers report that Opus 4.8 is more likely to flag uncertainties about its work and less likely to make unsupported claims. In internal evaluations, Opus 4.8 was about four times less likely than its predecessor to let flawed code pass unremarked.
Alignment assessments indicate new highs on prosocial traits such as supporting user autonomy and acting in the user’s best interest. Rates of misaligned behavior—e.g., deception—are lower than Opus 4.7 and comparable to the Claude Mythos Preview.
Benchmark Performance
- SWE‑Bench Pro: 69.2 % (outperforming GPT‑5.5 and Gemini 3.1 Pro)
- Terminal‑coding benchmark: GPT‑5.5 leads, but Opus 4.8 remains competitive on several other tests.
Opus 4.8’s fast mode runs at 2.5× the speed of previous versions and is three times cheaper to run.
New Product Features
- Dynamic workflows (research preview) – Claude can complete larger tasks in Claude Code, planning work and running hundreds of parallel sub‑agents in a single session. It can handle codebase‑scale migrations across hundreds of thousands of lines of code. Available for Claude Code Enterprise, Team, and Max plans.
- Effort control – In Claude.ai and Cowork, users can select the amount of effort Claude expends on a response. Lower settings yield faster replies and slower consumption of rate limits. Opus 4.8 defaults to high effort, which Anthropic says balances quality and user experience.
- Messages API – The Messages API now accepts system entries inside the
messagesarray, allowing developers to update Claude’s instructions mid‑task.
Availability and Pricing
Claude Opus 4.8 is available everywhere today. Pricing for regular use remains unchanged from Opus 4.7.
Future Roadmap
Anthropic is developing models that match Opus 4.8’s capabilities at a lower cost, as well as a new class of models that are even more intelligent. The company is also refining safeguards for the Claude Mythos model, currently being tested with a limited set of organizations, and expects to roll out Mythos‑class models to all customers “in the coming weeks.”