The advisor tool now supports a max_tokens parameter to cap the advisor model's output per call, reducing latency and output token cost for workloads that don't need full-length advisor responses. Set tools[].max_tokens on the advisor tool definition.
Fetched July 10, 2026
The @ai-sdk/anthropic package now forwards thinking: { type: 'disabled' } to the Anthropic API instead of silently dropping it, which was l…
Vercel · AI SDKCapturing of OpenAI pending background Responses without token usage is now skipped, reducing unnecessary data collection.
PostHog · PostHog JSWe've launched Claude Sonnet 5 (claude-sonnet-5), the next generation of our Sonnet model family, at introductory pricing of $2 / $10 per M…
Anthropic · Claude PlatformThe advisor tool now supports a max_tokens parameter to cap the advisor model's output per call, reducing latency and output token cost for…
Anthropic · Claude Platform