Grok 4 Fast

Grok 4 Fast is xAI’s cost- and latency-optimized tier of Grok 4, released September 19, 2025 for high-volume use cases like customer support, search, and agentic tool calling where running the full Grok 4 model on every request would be too slow or expensive. Despite the “fast” branding, it keeps a large 2-million-token context window and ships in both reasoning and non-reasoning variants so developers can pick the latency profile they need. Pricing is aggressive at $0.20 per million input tokens and $0.50 per million output tokens, undercutting most other frontier-adjacent models on a cost basis while still drawing on the same underlying Grok 4 training. xAI markets it heavily on agentic benchmarks and tool-use tasks rather than raw knowledge tests, and it’s available only through the API rather than the consumer Grok app or X.