llm-anthropic 0.26
Release: llm-anthropic 0.26 Includes new features enabled by LLM 0.32 : New models: claude-fable-5 , claude-sonnet-5 , and claude-opus-5 . #75 , #76 Added server-side tools for WebSearch , WebFetch , CodeExecution , and AnthropicMCP , available through LLM's -T interface or Python tools= . The previous -o web_search* options have been removed in favor of -T WebSearch . #79 Upgraded to llm>=0.32 .…
Anthropic has released a new version of their language models, with the release number 0.26. The updated models include claude-fable-5, claude-sonnet-5, and claude-opus-5. In addition, server-side tools such as WebSearch, WebFetch, CodeExecution, and AnthropicMCP have been added. These tools are now accessible through the LLM's -T interface or via the Python tools= parameter. The previous -o web_search* options have been replaced with the -T WebSearch option. The current version of the LLM has been upgraded to 0.32.
Anthropic has made several changes to improve the user experience. Reasoning, tool calls, tool results, and server-side tool results now stream as typed events. This allows for real-time feedback and interaction with the models. Additionally, reasoning for llm CLI prompts is now displayed on standard error, unless the --hide-reasoning/-R flag is used. This gives users more control over what information is displayed.
The extended thinking feature has been simplified, now referred to as "thinking" and "thinking_effort" with options of low, medium, high, xhigh, or max. By default, Claude 5 models engage in thinking. However, for Sonnet 5 and Opus 5, this can be disabled using the -o thinking 0 option. Fable 5 models always think by default. The --hide-reasoning/-R flag now omits reasoning from responses and logs, providing a cleaner output. Finally, the thinking_budget, thinking_display, and thinking_adaptive options have been removed.
Written by urgent.news from Simon Willison's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
