Grok 4 Fast Non-Reasoning is the stripped-down sibling in xAI's Grok 4 family, designed for high-throughput, low-cost inference rather than deep chain-of-thought reasoning. It is delivered as a proprietary, closed-weights model, so weights and full architecture details are not publicly available, but its intended role is clear from the pricing and benchmark posture: a fast, cheap workhorse that can sit behind large context windows for routine summarization, extraction, classification, and short-form generation workloads where the full Grok 4 would be overkill.

In practice, the model trades peak accuracy for affordability. Independent benchmark snapshots show it landing well below the flagship Grok 4 on math, science, and coding evaluations, scoring roughly a third on AIME 2025, around 62% on GPQA Diamond, and about 46% on LiveCodeBench, compared with the full model's 91.7%, 87.5%, and 79% respectively. The payoff is a blended cost roughly 96% lower than Grok 4, paired with a 2,000,000-token context window that makes it well suited to long-document processing, codebase ingestion, and bulk pipeline jobs where breadth and price matter more than squeezing out the last few accuracy points.