A New Flagship Goes Live
Anthropic positions Fable 5 as the newest and strongest, joining Opus 4.8, Sonnet 4.6, and Haiku 4.5 to make up the current Claude lineup; its model ID is claude-fable-5, and it leads with stronger reasoning and smoother output. As usual, the launch-day demos are always the best-looking part; what really matters is how it actually does on the tasks you run every day.
Don't Get Swayed by the Day-One Demo
What decides whether to switch has never been that it scores a few points higher on some benchmark, but whether the entry point is smooth, whether price and latency hold up, and whether the details—Chinese, long context, tool calling—all kept pace. Fable 5 and Claude 4.X aren't a matter of one replacing the other; it's more like tiering by difficulty and cost—hand the hard work to it, while the everyday light requests may not be worth it.
Whether to Switch—Run It Yourself
Don't rush to migrate your existing workflow wholesale. Do a small-sample comparison with real tasks: same batch of prompts, same set of files, same scoring, running Fable 5 side by side with your current default model, and note the success rate, time, and cost before deciding whether it enters the default chain or is only called in for hard problems.
via: Anthropic