Quick Take

This comparison evaluates xAI’s Grok Build 0.1 0616 against Xiaomi’s MiMo-V2.5. While Grok Build excels in coding-specific metrics and latency, MiMo-V2.5 provides a highly competitive intelligence index at a significantly lower price point.

Benchmark Read

Performance metrics highlight different strengths for each model. Grok Build 0.1 0616 leads in coding (51.5) and shows strong results in GPQA (0.895), HLE (0.36), SciCode (0.502), and LCR (0.646). MiMo-V2.5, released April 22, 2026, holds a slightly higher overall intelligence index of 40.1. It demonstrates robust capabilities in specialized benchmarks, including TAU2 (0.906), IFBench (0.671), and TerminalBench Hard (0.416), though it trails Grok in GPQA (0.849), HLE (0.252), and SciCode (0.431).

Cost and Speed

Cost is a major differentiator. Grok Build 0.1 0616 is priced at a blended rate of $1.25/1M tokens, with input at $1.00 and output at $2.00. MiMo-V2.5 is significantly more affordable, with a blended rate of $0.17/1M tokens, featuring input at $0.14 and output at $0.28. Regarding speed, Grok Build offers a faster time to first token (0.405s) compared to MiMo-V2.5 (2.195s), though both models maintain similar output speeds of approximately 70 tokens per second.

Best Fit

Grok Build 0.1 0616 is best suited for developers requiring a responsive, coding-centric agent. Its integration with terminal-based workflows makes it ideal for technical tasks. MiMo-V2.5 is best suited for high-volume enterprise applications where cost-efficiency and performance across a wider array of logic and instruction-following benchmarks are prioritized.