Developer releases Anypick, open-source library for filtering LLMs by capabilities and price
A library, implemented in both Python and TypeScript, that downloads model catalogs (e.g. from OpenRouter or Vercel) and lets developers build pipelines to filter models by price, latency, benchmarks, and capabilities, then pick the best one by criteria like price or throughput. It targets developers building LLM systems who want to swap or update underlying models without rewriting code every few months. The LLM connection is that it operates on LLM catalogs and metadata rather than calling models itself.
A library, implemented in both Python and TypeScript, that downloads model catalogs (e.g. from OpenRouter or Vercel) and lets developers build pipelines to filter models by price, latency, benchmarks, and capabilities, then pick the best one by criteria like price or throughput. It targets developers building LLM systems who want to swap or update underlying models without rewriting code every few months. The LLM connection is that it operates on LLM catalogs and metadata rather than calling models itself.
Anansi is an open-source memory API designed to give LLM applications persistent, queryable memory. It solves the problem of stateless LLM interactions by providing a simple interface for storing and retrieving conversation context or knowledge, aimed at developers building LLM-powered apps.
Anansi is an open-source memory API designed to give LLM applications persistent, queryable memory. It solves the problem of stateless LLM interactions by providing a simple interface for storing and retrieving conversation context or knowledge, aimed at developers building LLM-powered apps.
RAGless is a system that provides retrieval-augmented generation-like functionality without incurring LLM API costs at runtime. It likely uses local or precomputed methods to generate responses, targeting developers who want to reduce operational expenses. The project is open-source and hosted on GitHub.
RAGless is a system that provides retrieval-augmented generation-like functionality without incurring LLM API costs at runtime. It likely uses local or precomputed methods to generate responses, targeting developers who want to reduce operational expenses. The project is open-source and hosted on GitHub.
Parley is a self-hosted mesh network that connects LLMs running on your own devices, allowing them to collaborate on tasks without relying on external APIs. It solves the problem of using multiple local models together efficiently, for users who want privacy and control over their AI workflows.
Parley is a self-hosted mesh network that connects LLMs running on your own devices, allowing them to collaborate on tasks without relying on external APIs. It solves the problem of using multiple local models together efficiently, for users who want privacy and control over their AI workflows.
Philipp Dubach launches Model Atlas, an LLM leaderboard with user-reweightable metrics
Philipp Dubach launches Model Atlas, an LLM leaderboard with user-reweightable metrics