Baseten ships deferred tool loading: agents run up to 62% cheaper and 26% faster

baseten · x · 2026-11-20

Baseten announced that deferred tool loading is now live for DeepSeek, Kimi, and GLM on its Model APIs. The companion article argues agent harness building has evolved from addition to subtraction: cutting prompt bloat from tool definitions pays off.

Key points:

Original post →

More from coding & agent

coding & agent channel →