Anthropic Opus 4.8 with ‘better’ honesty and coding abilities launched: All details

Anthropic has launched Claude Opus 4.8, an upgrade to its flagship artificial intelligence (AI) model. The company says that the new version brings significant improvements in coding, reasoning and practical knowledge tasks. It is available immediately to users for the same price as its predecessor. Alongside the model, Anthropic has rolled out a suite of cost-saving tools and features aimed at giving users more control over how the AI tackles large-scale problems.A common issue among advanced AI models is their tendency to jump to conclusions, confidently claiming they have solved a problem even when the evidence is thin. Anthropic claims to have made a major breakthrough in fixing this with Opus 4.8.Citing early testers, Anthropic says that Opus 4.8 is significantly better at flagging uncertainties in its own work and is much less likely to make unsupported claims. Meanwhile, Anthropic's internal evaluations found that the model is about four times less likely than its predecessor, Opus 4.7, to let flaws in its written code pass by unremarked.The model is also claimed to have outperformed its competitors on several key industry benchmarks, specifically excelling in financial analysis, reasoning and agentic coding.As businesses look for more budget-friendly ways to deploy AI, Anthropic's latest launch emphasises cost efficiency and usage customisation.Cheaper ‘Fast Mode’: Opus 4.8 features a "fast mode" that allows the AI to work at 2.5 times its normal speed, the company claims, adding that this mode is now three times cheaper to run than it was on previous iterations.User effort controls: Users on Claude.ai can now manually dictate how much “effort” the AI puts into a response. High-effort settings will cause the model to “think” longer for complex tasks and vice-versa, helping them save on token limits.Dynamic Workflows for massive projects: Third feature is for software developers, the “Claude Code” platform introduces a “dynamic workflows” feature, allowing the AI to spin up and run multiple subagents at once to break down and solve massive programming problems.Anthropic claims that the model showed a substantial drop in harmful or misaligned behaviours, such as deception or cooperating with misuse.