Anthropic today announced the launch of its latest AI model, Claude Opus 4.8. Anthropic claims the model is a “more effective collaborator” with improvements in agent coding, multidisciplinary reasoning, agent computer use, knowledge work, and agent financial analysis.
Testers found Opus 4.8 to be “more reliable and accurate in its judgment” when performing agentic tasks, and the model also gained in honesty.
Early testers report that Opus 4.8 is more likely to report uncertainties about its work and less likely to make unsupported claims. This is confirmed by our evaluations, which show that Opus 4.8 is about four times less likely than its predecessor to leak flaws in the code it wrote.
Alignment assessments suggest that the model is reaching new heights in terms of measures of prosocial traits such as supporting user autonomy and acting in the user’s best interest. Rates of misaligned behaviors like deception are lower than Opus 4.7 and similar to Claude Mythos Preview.
Anthropic benchmarks indicate that Opus 4.8 scored 69.2% on SWE-Bench Pro, outperforming GPT-5.5 and Gemini 3.1 Pro on the test and several other benchmarks, although GPT-5.5 led on the terminal encoding benchmark.
The Opus 4.8’s fast mode also works at 2.5 times the speed and is now three times cheaper than previous models.
Alongside Opus 4.8, Anthropic is adding new features to its product line.
- Dynamic workflows (search overview) – Claude can perform more important tasks in Claude Code. It is capable of scheduling work and executing hundreds of parallel subagents in a single session. It is capable of performing codebase-wide migrations across hundreds of thousands of lines of code. The feature is available for the Claude Code for Business, Team and Max plans.
- Effort control – In Claude.ai and Cowork, users can choose how much effort Claude puts into a response. At a lower setting, Claude will respond faster and use rate limits more slowly. Opus 4.8 defaults to high effort, which Anthropic says is the best balance between quality and user experience.
- Messages API – The Messages API accepts system entries in the messages table, so developers can update Claude’s instructions mid-task.
Claude Opus 4.8 is available everywhere today. The price for regular use has not changed compared to Opus 4.7.
Anthropic is working on models with the same capabilities as Opus 4.8 at a lower cost, as well as a new class of models that are even smarter than Opus. Anthropic says it has developed backups for the Claude Mythos model which it is testing with a small number of organizations and hopes to be able to offer Mythos class models to all customers “in the coming weeks.”
