LLM Reference

LLM Reference is a directory and comparison site for language models, providers, and benchmarks. It is built for developers, knowledge workers, and teams that need to choose a model for tasks like coding, agents, writing, research, data and SQL, image generation, video, voice, transcription, and music.

The site centers on helping users pick the right model for a specific job rather than browsing a generic catalog. It offers search across the model directory, curated picks by use case, and side-by-side comparison of models and providers. The homepage also highlights weekly market changes such as new models, price cuts, and benchmark refreshes, so users can keep up with a fast-changing field.

LLM Reference presents board-style leaderboards for different audiences and tasks, including developers, knowledge workers, and creatives. It also surfaces “best overall today” and “freshest update” selections, plus a live shortlist that reflects the current default view of coding task, balanced budget, and fresh research. For teams evaluating options, the site’s benchmark-driven approach is designed to make it easier to compare quality, cost, and recency before choosing a model or provider.

It also includes tools and references such as a model API, compare pages, benchmarks, frontier pricing, changelog, methodology, and a search index. The site is updated daily and states that it tracks 1,897 language models, 143 providers, and 249 labs.

Key features:
- Searchable directory of language models and providers
- Comparison tools for two models and model/provider options
- Curated picks for coding, agents, writing, research, and creative tasks
- Leaderboards organized by audience and use case
- Weekly pulse updates for new models, price cuts, and benchmark refreshes
- Frontier pricing and benchmark tracking across many suites
- API, methodology, changelog, and search index pages for deeper review