3 articles tagged
Anthropic's browser use tool went GA on 19 August 2026. Declaring it costs about 6,600 input tokens per request against 4,500 for computer use, and its own docs contradict the release notes.
Editing the tools array mid-session invalidates the entire Claude prompt cache. The mid-conversation-tool-changes-2026-07-01 beta replaces that edit with two content blocks, and the price gap between a cache read and
State-of-the-art function-calling agents still solve under 50% of real tasks and stumble on retries. A 2026 guide to evaluating LLM tool-use reliability with pass^k, tau-bench, BFCL, and your own eval harness.
