Source details
- Original source
- Towards AI
- Published
- 2026-09-25
- Primary topic
- Foundation Models
Why it matters
Model launches, benchmark jumps, API upgrades, context window changes, and frontier LLM competition. Use the original source for the full report, then use the directory shortcuts below to compare the products and workflows the story points toward.
What happened
Author(s): Chew Loong Nian – AI ENGINEER Originally published on Towards AI. Anthropic’s SDK 0.128.0, released with Claude Opus 5.5, can hand the model a new tool mid-run without touching tools[]. It needs one beta flag the runner won’t add for you. If your Claude agent picks up a new tool halfway through a long conversation, the way you add it decides how much of the next request can still come from the prompt cache: 98.7% of it with the new runner.addTools(), none if you edit tools[]. The article explains what changed in Anthropic’s SDK 0.128.0: a new inline tool runner supports addTools() and removeTools(), allowing tools to be added during an active conversation. It contrasts this with the older approach of editing tools[], which is expensive because tool definitions sit at the front of the cached prompt prefix—so changing them invalidates the entire cache. With addTools(), the existing cached prefix remains intact and the new tool is sent as an appended system message, yielding very high request reuse (measured up to 98.7% on a 40-turn conversation). It provides a no-API-key Node.js script to measure request-body reuse, shows how the cache-reuse gap grows as conversation length increases, and calls out a key gotcha: you must explicitly include the inline-tools-2026-09-15 beta in the runner parameters because the SDK will not add it automatically. Finally, it notes which models and platforms support mid-conversation tool changes, plus several practical details (edge cases that can cause full misses, how pause turns delay the change, and that add/remove actions take effect immediately on the next step), ending with a recommendation to keep tools[] unchanged and always add the required beta flag. Read the full blog for free on Medium. Join thousands of data leaders on the AI newsletter. Join over 80,000 subscribers and keep up to date with the latest developments in AI. From research to projects and ideas. If you are building an AI startup, an AI-related product, or a service, we invite you to consider becoming a sponsor. Published via Towards AI
What to do next
Compare the hosted model pages first, then check the related tools and buyer guides before changing workflow standards.
Author(s): Chew Loong Nian – AI ENGINEER Originally published on Towards AI. Anthropic’s SDK 0.128.0, released with Claude Opus 5.5, can hand the model a new tool mid-run without touching tools[]. It needs one beta flag the runner won’t add for you. If your Claude agent picks up a new tool halfway through a long conversation, the way you add it decides how much of the next request can still come from the prompt cache: 98.7% of it with the new runner.addTools(), none if you edit tools[]. The article explains what changed in Anthropic’s SDK 0.128.0: a new inline tool runner supports addTools() and removeTools(), allowing tools to be added during an active conversation. It contrasts this with the older approach of editing tools[], which is expensive because tool definitions sit at the front of the cached prompt prefix—so changing them invalidates the entire cache. With addTools(), the existing cached prefix remains intact and the new tool is sent as an appended system message, yielding very high request reuse (measured up to 98.7% on a 40-turn conversation). It provides a no-API-key Node.js script to measure request-body reuse, shows how the cache-reuse gap grows as conversation length increases, and calls out a key gotcha: you must explicitly include the inline-tools-2026-09-15 beta in the runner parameters because the SDK will not add it automatically. Finally, it notes which models and platforms support mid-conversation tool changes, plus several practical details (edge cases that can cause full misses, how pause turns delay the change, and that add/remove actions take effect immediately on the next step), ending with a recommendation to keep tools[] unchanged and always add the required beta flag. Read the full blog for free on Medium. Join thousands of data leaders on the AI newsletter. Join over 80,000 subscribers and keep up to date with the latest developments in AI. From research to projects and ideas. If you are building an AI startup, an AI-related product, or a service, we invite you to consider becoming a sponsor. Published via Towards AI
This AimostAll brief summarizes the linked source so readers can scan AI developments quickly and jump to the original reporting when needed.
