Very neat, bookmarked. Read through the readme.md and would be nice to show an example of using with Ollama or LM Studio running locally. Would guess you just setup an OpenAI model, set the baseUrl to the local instance / API and go from there? For something that is Fully Local connecting to a provider with an API key is iffy at best, unless you are paying for enterprise keys. I guess the first line says used by ‘Enterprises’ but to be fully local that would include running the LLM locally also?
That is great. How do you even handle the daily divergence that must happen due to micro decisions made by various APIs. For example I wont mention their name but one OpenAI compatiable provider doesn't fully work with OpenAI client and had various holes so you need to account for that. The other hard thing is just the fast pace of AI. What if the underlying concepts change!
A more fleshed out competitor to llm-interface [1]? Looks great!
Does it run in FaaS/serverless environments out of the box? Lambdas, Cloudflare Workers, Vercel functions and the likes? Deno? The README says "Isomorphic - works everywhere", but might be nice to make this more explicit.
The fully local production-grade Super SDK that provides a simple, unified, and powerful interface for calling more than 200+ LLMs.
- Production-ready and used by enterprises.
- Fully local and NOT a proxy. You can deploy it anywhere.
- Comes with batching, retries, caching, callbacks, and OpenTelemetry support.
- Supports custom plugins for caching, logging, HTTP client, and more. You can use it like LEGOs and make it work with your infrastructure.
- Supports plug-and-play providers. You can run fully custom providers and still leverage all the benefits of Adaline Gateway.
Features
Strongly typed in TypeScript
Isomorphic - works everywhere
100% local and private and NOT a proxy
Tool calling support across all compatible LLMs
Batching for all requests with custom queue support
Automatic retries with exponential backoff
Caching with custom cache plug-in support
Callbacks for full custom instrumentation and hooks
OpenTelemetry to plug tracing into your existing infrastructure
Plug-and-play custom providers for local and custom models
Looks good, might give it a try. I was looking for something similar to provide a unified interface for gpt and claude and eventually hacked something together myself, as none of the solutions I found could deal with structured output properly across both vendors.
I think it's because the library calls directly the LLM AND saves any debug/trace info. While other tools let you use the standard (eg OpenAI) sdk and uses an https proxy to intercept the requests
We have implemented “super-types” that are strongly typed in TypeScript and compatible with all providers (think a superset in mathematics). Hence the name, Super SDK. Gateway is super light weight. We don’t wrap around other SDKs. We describe the APIs in the provider packages (@adaline/openai) that plugs into @adaline/gateway that has all the features.
