LMSpeed is a platform for researching and comparing large language models and the APIs that provide access to them. It brings together model benchmarks, provider pricing, performance testing, and security assessments to help users evaluate options before choosing an API. The model directory presents scores and metadata across areas such as overall intelligence, coding, reasoning, mathematics, and other capabilities, alongside details including context limits, token rates, latency, and throughput where available.

LMSpeed compares input and output token pricing across more than 100 providers and highlights free tiers and credits, helping users identify lower-cost options for a given model. Its tools include model benchmark lookup, real-time speed tests using standardized prompts, API health checks, and analytics for latency, throughput, and response duration. Users can also benchmark custom endpoints, including official providers, proxies, relays, and self-hosted services. Security audits for OpenAI-compatible APIs assess model identity, prompt safety, instruction tampering, response and stream integrity, and error leakage. Audit reports and leaderboards make results easier to review and share.