xAI SDK Fix Recovers Lost Reasoning Tokens, Boosts Grok Ranking
- Source
- x.com
- Date
Post-launch, xAI has updated its SDK, significantly improving its performance. It is now #10 on the Vals Index.
Context
Vals AI reports that after launch, xAI updated its SDK in a way that meaningfully improved measured performance, moving xAI up to #10 on the Vals Index, an independent of coding and agentic performance. The post itself doesn't explain the underlying bug.
Separate technical reports from around the same period describe a matching problem: xAI's API returns each reasoning turn as an encrypted block, and Grok models need that block sent back on the next request to keep reasoning intact across a multistep tool-use loop; when client SDKs dropped it, Grok's reasoning could stop working after about two tool calls. That mechanism is consistent with, but not directly confirmed by, Vals AI's post as the specific fix behind the ranking change.
- benchmark — A standard public test set for comparing AI models — the shared scoreboards behind every "model X beats model Y" claim.
- context window — The maximum amount of text a model can consider at once — its working memory for the current conversation or task.
Checking sign-in…
Loading comments…