Recommended Free Tools
Chew Loong Nian’s September 25, 2026, Towards AI article reports that adding a tool with Claude’s inline runner addTools() reused up to 98.7% of the next request body in one 40-turn conversation; editing the request’s tools[] array produced no reuse in the same test. That is a reported, workload-specific result—not an Anthropic guarantee, a universal cache rate, or proof of lower latency or API costs.
What the 98.7% figure measures
The percentage is the author’s maximum reported request-body reuse after adding a tool partway through a 40-turn conversation. In the article’s comparison, changing tools[] yielded no reuse for the next request. The result is from one reported test; no multi-run study or independent reproduction is established by the sources available here.
As an Amazon Associate I earn from qualifying purchases.
Request-body reuse should not be conflated with a cache hit rate across all workloads, faster responses, fewer billed tokens, or a specific cost reduction. The article does not provide sourced measurements for those outcomes.
How the two approaches differ in the report
| Approach | How the article describes the change | Reported next-request reuse |
|---|---|---|
Inline runner addTools() |
Append a tool addition while a run is in progress, preserving the earlier prompt prefix, according to the article. | Up to 98.7% in the author’s 40-turn test. |
Edit the request’s tools[] |
Change the tool array, which the article says alters the cached prompt prefix. | No reuse in the same reported comparison. |
The explanation about prefix preservation is the author’s account of the mechanism; the official Anthropic documentation located for this article does not independently confirm how these methods work. Nor does the single comparison establish that every direct tool-list edit invalidates all reusable content in every API setup.
#1 Best Overall
What the article says about addTools() and beta support
The Towards AI article describes Anthropic SDK 0.128.0 as including an inline tool runner with addTools() and removeTools() methods for changing available tools during a run. It also says the runner does not add the beta flag automatically: developers must explicitly include inline-tools-2026-09-15.
These are version-sensitive claims from secondary reporting. The Anthropic prompt-engineering page provides general context on tools and prompting, but does not establish these methods, the beta flag, or the reported benchmark: Anthropic prompt-engineering documentation. Confirm current SDK documentation, beta requirements, and model and platform support before relying on the feature in a production integration.
Rank #2
How to choose an implementation
- Use the reported result as a reason to investigate, not a performance guarantee. If tools need to change during a long-running interaction, check whether the current SDK and model environment support the inline runner.
- Verify behavior in your actual request flow. Compare the runner approach with updating
tools[]under the same conversation, model, and workload, and measure the reuse signal you care about. - Measure operational outcomes separately. Track API billing and end-to-end latency directly; the article’s request-body reuse percentage does not establish either one.
- Check the beta requirement before shipping. Confirm the current flag name and how to pass it in your SDK and platform rather than assuming the article’s September 2026 guidance remains current.
The article also reportedly includes a Node.js script intended to measure request-body reuse without an API key. Its presence is not independent verification of the result, and no execution or reproduction of that script is established here.
The Tool Desk
Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →What is independently established
The feature description, mechanism explanation, explicit beta-flag guidance, and 98.7% result here all trace to Chew Loong Nian’s September 25, 2026, Towards AI article. The official Anthropic source cited above is general prompt-engineering guidance, not confirmation of the feature or benchmark. Treat the number as an attributed report until current official documentation and measurements in your own environment support a stronger conclusion.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
