[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"$f1n570u109789n":3},{"code":4,"message":5,"data":6,"timestamp":39},10000,"success",{"full_name":7,"position":8,"description":9,"language":10,"stars":11,"forks":12,"stars_gained":13,"age_days":14,"hn_threads":15,"one_liner":16,"summary_md":17,"topic":18,"github_topics":21,"license":29,"homepage":30,"updated_at":31,"history":32},"QuantumNous\u002Fnew-api",14,"A unified AI model hub for aggregation & distribution. It supports cross-converting various LLMs into OpenAI-compatible, Claude-compatible, or Gemini-compatible formats. A centralized gateway for personal and enterprise model management.","Go",50648,12248,51,1079,[],"A self-hosted gateway that puts many model providers behind one OpenAI-, Claude- or Gemini-compatible API for teams and agents.","If your team already juggles several model accounts, New API is worth a look before you hand clients a pile of provider keys. It is a self-hosted AI gateway you deploy yourself, not a hosted service.\n\n## The problem\n\nA team usually starts with per-provider keys scattered across tools: one client is configured for OpenAI, another for Anthropic, a third for Gemini or a local provider. Every new application means redoing credential setup, and nobody has one place to see who spent what. New API changes that step: you connect upstream providers once, expose a single endpoint family, and let clients point at the gateway instead of each vendor. The result is one place to route requests, issue client keys and read usage and cost, rather than reconciling several dashboards.\n\n## What's different\n\nIts distinguishing mechanism is protocol conversion in front of the upstreams. The same deployment accepts OpenAI Chat Completions and Responses, Anthropic Messages and Gemini generateContent\u002FstreamGenerateContent, including streaming, tools, reasoning and multimodal inputs where the provider supports them. That means a Claude-shaped client can talk to a Gemini-backed channel without you rewriting the client. Requests then flow through configurable routing: model mappings, channel priorities and weights, retries, channel affinity and multiple upstream keys per channel.\n\nThat is a meaningful operational shift from simply proxying one API. Routing and retry rules live centrally, so adding a provider does not mean reconfiguring every consumer. Usage and cost tracking, quotas, subscriptions, cache accounting and expression-based per-tier pricing are handled in the same console.\n\nThe trade-off is that you now run a critical-path service. Conversions and routing are extra moving parts: a model may behave differently when reached through a compatibility layer, and the gateway becomes a place to monitor and upgrade. It also only makes sense when you legitimately hold upstream access you are allowed to redistribute internally.\n\n## Who should try it\n\nIt fits a team or small company that holds authorized access to several providers and wants internal apps and agents to share it through one endpoint with centralized quotas and logs. It also suits someone running a private multi-model setup who prefers a web console over editing config files per client.\n\nIt is a poor fit if you are a single developer with one provider key and no sharing requirement: the extra hop and admin surface buy you little. It is also the wrong tool for anyone expecting the project to supply model access rather than route access they already have.\n\n## Requirements and limitations\n\nThe README's local path is a Docker container with SQLite, bound to localhost, with a mounted data directory for persistence. For anything beyond a local trial you must plan the storage and deployment yourself; the sample is a starting point, not a production topology. The project is licensed AGPL-3.0 with additional Section 7 terms: modified versions must keep an author attribution notice and a visible link to the original repository. Organizations that disallow AGPL obligations are pointed to commercial contact options.\n\nOngoing effort is real: upstream providers, keys, model mappings and pricing expressions all need maintenance, and the gateway sits in front of every request, so its upgrades and availability matter.\n\n## HotRepo's take\n\nHotRepo's dated snapshots show 50,622 stars on 2026-10-24 and 50,630 on 2026-10-25, with two days on the daily board. That two-day sample is too short to judge whether interest is rising or fading. Before committing, read the [project documentation](https:\u002F\u002Fwww.newapi.ai) for the exact protocol-conversion caveats, and confirm your upstream providers permit the internal redistribution you plan.",{"slug":19,"name":20},"ai-engineering","AI Engineering",[22,23,24,25,26,27,28],"ai-gateway","claude","deepseek","gemini","newapi","openai","rerank","AGPL-3.0","https:\u002F\u002Fwww.newapi.ai","2026-10-25T02:59:08Z",[33,37],{"date":34,"stars":35,"stars_gained":36},"2026-10-24",50622,52,{"date":38,"stars":11,"stars_gained":13},"2026-10-25",1792921175]