GMH / Research Log
Engineering notes for teams building with multiple AI model providers—covering routing, cost, latency, reliability, and integration.
Routing Reliability Cost IntegrationA practical guide to LLM routing strategies: what cost, latency, availability, task, budget, and rule based routing each solve, how static and dynamic routing differ, and how to combine signals into one decision.
Read diagnosis
Research archive
Technical writing on routing decisions, provider behavior, API design, latency, cost, and reliability.
An educational, developer-focused guide to Time to First Token (TTFT): what it measures, why it matters for streaming LLM apps, which factors affect it, and…
A developer-focused, evidence-based comparison of Claude vs Gemini across the dimensions that matter for real model selection.
A developer-focused guide to LLM inference and API performance metrics, from TTFT to throughput and latency percentiles.
A developer-focused guide to LLM latency: what it is, where the time goes in a single request, how to measure it, and the engineering and…