Repository navigation
Proposal: UberLLM as a provider for chats and transcriptions #1197
al-global-tech
started this conversation in
Proposals
Replies: 0 comments
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
Problem, from the user's side. Some AI Studio users want one prepaid account for many chat models (Claude, GPT, Gemini, DeepSeek, Qwen, GLM, Kimi and others) without a monthly plan, and want to pay less than list price. Today they can only reach such a service through the self-hosted/OpenAI-compatible provider, which gives them no model list metadata, no key link, and no confidence information.
Intended solution. Add UberLLM as a provider the same way Requesty was added in #1006: a
ProviderUberLLMonhttps://api.uberllm.dev/v1/(OpenAI-compatible chat completions with streaming), a host that splits itsvendor/modelids like the OpenRouter host, the model list fromGET /v1/models(it reportscontext_lengthandmax_output_tokens), the attribution headers fromProjectAttribution, and transcription through/v1/audio/transcriptions(six speech-to-text models today). UberLLM has no embedding models, so embeddings would stay off.What you should know for the confidence level. UberLLM is a US service (operated from Wyoming). Each request is sent to a third-party provider that serves the model (for example OpenRouter, Together AI, Fireworks AI, DeepSeek, Amazon Bedrock); how providers are checked is at https://uberllm.dev/verification. There is no EU endpoint or EU-only routing, so it would not fit organizations that need EU processing; I'd expect a confidence level like the other US gateways, and I'm glad to follow your lead on that.
If this is a "Go", I'll mirror #1006 (provider, host naming, model list metadata, icon with credit, i18n entries, changelog, tests) and run
dotnet run verifybefore opening the PR. Prices per model: https://uberllm.dev/modelsI work on UberLLM.
All reactions