Grok 4.20 Non-Reasoning is xAI's non-reasoning model in the Grok 4.20 beta generation, optimized for speed and direct responses with low hallucination rates and strict prompt adherence.
import { streamText } from 'ai'
const result = streamText({ model: 'xai/grok-4.20-non-reasoning', prompt: 'Why is the sky blue?'})Grok 4.20 Non-Reasoning was released March 9, 2026 as part of xAI's Grok 4.20 beta generation. It's optimized for speed and direct responses, producing answers without chain-of-thought reasoning overhead. The model features low hallucination rates and strict prompt adherence, making it suitable for production workloads that need precise, reliable output.
As a non-reasoning variant, Grok 4.20 Non-Reasoning skips intermediate reasoning traces and delivers answers directly. This reduces latency and output token cost per request, which is particularly valuable in high-throughput applications and agentic tool-calling loops where per-step speed compounds into overall workflow efficiency.
This model is currently in beta.