Relace Blog
Relace Compact: 50k TPS for 50% cost savings
Relace Compact: 50k TPS for 50% cost savings
Relace Compact: 50k TPS for 50% cost savings
Reducing token spend on frontier models with fast compaction on cache-miss.
Jacq is now available on macOS. Download and try it for free at
Try Jacq for free at
Models
Relace Compact: 50k TPS for 50% cost savings
Published Jul 13, 2026
Reducing token spend on frontier models with fast compaction on cache-miss.
Models
Relace Compact: 50k TPS for 50% cost savings
July 13, 2026
Reducing token spend on frontier models with fast compaction on cache-miss.
Models
Exploiting parallel tool calls to make agentic search 4x faster
Published Dec 8, 2025
Today we're releasing Fast Agentic Search (FAS), a code-specific subagent trained with RL to search through codebases for files relevant to a user request.
Models
Exploiting parallel tool calls to make agentic search 4x faster
December 8, 2025
Today we're releasing Fast Agentic Search (FAS), a code-specific subagent trained with RL to search through codebases for files relevant to a user request.
Models
A Year of Fast Apply — The Path to 10k Tokens per Second
Published Oct 29, 2025
Relace Apply 3, and how we built the series of models that took us past 1M ARR.
Models
A Year of Fast Apply — The Path to 10k Tokens per Second
October 29, 2025
Relace Apply 3, and how we built the series of models that took us past 1M ARR.