NewPoint any agent at UberLLM — one key, every model, list price or less.
UberLLM

Use UberLLM with SillyTavern

SillyTavern connects to UberLLM through its Custom (OpenAI-compatible) chat completion source. This guide is kept by UberLLM, following SillyTavern's guide for compatible providers.

Prerequisites

  • An UberLLM account with credit
  • An API key from Dashboard → Keys (ull_...)
  • A model id, for example deepseek/deepseek-v4.1-flash

Connect

  1. In SillyTavern, open API Connections (the plug icon).
  2. Set API to Chat Completion.
  3. Set Chat Completion Source to Custom (OpenAI-compatible).
  4. In Custom Endpoint (Base URL), enter https://api.uberllm.dev/v1. Use the API root ending in /v1, not /chat/completions: SillyTavern adds that itself.
  5. In Custom API Key, paste your key.
  6. Click Connect. Pick a model from Available Models (loaded from /v1/models), or type its id in Enter a Model ID.
  7. Prompt Post-Processing: leave it on None. If a model rejects your prompt layout (for example, two messages in a row from the same role), try Merge consecutive roles.
  8. Click Test Message.

SillyTavern's server makes the calls, not your browser, so nothing else is needed.

Recommended models

Checked on 2026-10-30 through UberLLM.

Model id Context Notes
deepseek/deepseek-v4.1-flash 1,048,576 Fast and cheap; reasons before it answers
z-ai/glm-5.2 1,048,576 Good at long stories
venice/uncensored-1.2 128,000 Uncensored
deepseek/deepseek-v4.1-flash-uncensored 1,048,576 Uncensored; fewer providers serve it, so it can be busy

Uncensored text models are listed for every account. Prices change with the provider serving each request: see the model pages.

Feature support

  • Model discovery through /v1/models: yes, with your key. The dropdown lists chat models.
  • Streaming: yes.
  • Tool calling: yes, on models that support tools.
  • Vision: yes, on models that read images (for example google/gemini-2.5-flash).
  • Reasoning: reasoning models stream their reasoning in reasoning_content (and the same text in reasoning) before the answer, whichever provider serves them.
  • Structured output: yes (response_format with a JSON schema), on models that support it.

Known limitations

  • Each request goes to a provider that serves the model. Answers to the same prompt can differ a little between providers.
  • Additional Parameters → Include Body Parameters passes extra fields to the provider as they are. Fields a model does not know may be ignored or refused.
  • Not tested: text completion mode, and SillyTavern's image generation through this source.

Common errors

What you see Cause and fix
Connect fails, 401 Key missing or mistyped.
404 "page not found" The endpoint has /chat/completions at the end, or a typo. Use https://api.uberllm.dev/v1.
404 model_not_found Model id mistyped. Pick it from Available Models.
402 insufficient_credits Balance empty. Add credit under Dashboard → Credits.
503 no_available_provider Nobody can serve that model right now. Retry or pick another model; nothing was charged.

Support

For account, billing, API or model problems, write to [email protected].

Last verified with SillyTavern 1.19.0 (staging, 2026-10-03) on 2026-10-30.