Grok 4 Fast (Non-Reasoning)
Release Date: 9/19/2025
Benchmarked by 
Cost-efficient model focused on speed and efficiency for straightforward tasks like summarization or classification without deep logical processing. Unified architecture with reasoning variant, steered via system prompts.
Avg. Accuracy
48.4%
Latency
53.0s
Cost (In/Out)
0.20 / 0.50
Context Window
2M
Max Output Tokens
N/A
Input Modality
Benchmarks
Accuracy
Rankings
Academic Benchmarks
Proprietary Benchmarks (contact us to get access)