Today at 11:05 AM

xAI has released Grok 4 Fast, the company’s latest AI model, advancing cost-efficient reasoning for artificial intelligence. Building on lessons from Grok 4, Grok 4 Fast delivers frontier-level performance for both enterprise and consumer domains, while emphasizing outstanding token efficiency for end users and developers.

Key improvements include a 2 million token context window, cutting-edge web and X search capabilities, and a unified model architecture. Grok 4 Fast introduces a system that handles both long chain-of-thought reasoning and quick responses within the same set of model weights, guided by system prompts. This unification removes the previous need for separate models, reducing end-to-end response latency and token use. These architecture changes particularly benefit real-time applications seeking lower latency and higher throughput.

Compared to Grok 3 Mini, Grok 4 Fast achieves better scores on reasoning benchmarks while delivering a significant reduction in token costs. When matched against Grok 4, the model maintains comparable performance on frontier benchmarks but achieves this with 40% fewer tokens. The combination of improved token efficiency and a lower per-token price drives a 98% cost reduction for users requiring high-level reasoning.

Additionally, Grok 4 Fast was trained end-to-end using tool-use reinforcement learning, optimizing when to leverage code execution and web browsing. The new AI model is now generally available to all users via the website and iOS/Android mobile app.

Today by Paul