Merge Gateway Targets Coding-Agent Spend

Merge is betting that AI cost control belongs below the application layer, especially as coding agents generate high-volume, multi-model workloads.

• –Intelligent routing can trade model quality against latency and cost per request
• –Project-level budgets and hard limits help prevent runaway agent spending
• –Automatic failover reduces dependence on any single model provider
• –Request-level logs expose routing decisions, costs, latency, and policy outcomes
• –Context compression and semantic caching target the token waste common in agentic workflows