Replace your ChatGPT, Claude, or Gemini subscription on iPhone, iPad, and desktop. CodeBurst exposes a standard OpenAI-compatible API, so any chat client that accepts a custom base URL + API key works out of the box — and you get multi-model chains instead of a single model.
In whichever app you choose, you'll be asked for three things:
| API base URL | https://codeburst.ai/v1 |
| API key | Your CodeBurst key (43 random characters) — get one at /api-keys |
| Model name | Any CodeBurst model: codeburst-free, codeburst-smart, or a custom chain like grug-triple-check |
https://codeburst.ai/v1codeburst-free (or any other tier)/v1/responses and "browse models" silently fails to populate. Turning it off makes OpenCat use /v1/chat/completions instead, which works out of the box — model browse populates and messages send normally.
Verified working: model browse populates, streaming responses arrive, codeburst-free returns answers in 1-3s via Gemini Flash / Cohere / Groq fallback chain.
https://codeburst.ai/v1https://codeburst.ai/v1codeburst-smart (or any other)| Pal Chat skip | Limited to ~10 baked-in providers, no custom endpoint support. |
| ChatGPT app skip | Locked to OpenAI's servers, cannot point at a custom backend. |
| Claude app skip | Locked to Anthropic. |
| codeburst-free | Cycles free-tier providers (Gemini Flash / Cohere / Groq). No looping. Free. |
| codeburst-cost-saver | Free models first, falls back to cheap paid models. Mostly free. |
| codeburst-fast | Single ultra-fast tiny model. Cheap. |
| codeburst-smart | High-quality single-pass code generation. Standard. |
| codeburst-heavy | Premium reasoning models. Higher cost. |
| grug-triple-check | Custom chain (live). Draft → critique → revise, 3 passes. Higher quality for important questions. |
| grug-deep-think | Custom chain (live). Reason step-by-step → self-verify, 2 passes. |
Custom chains follow the {username}-{chainname} convention. Build them visually in CodeBurst Studio — the multi-model AI playground. Once you
define a chain (e.g. grug-deep-think,
grug-brainstorm-committee), it becomes callable from every app on this
page using just its name.
The base URL must be https://codeburst.ai/v1 exactly. Some apps
auto-append /v1, so try just https://codeburst.ai
if the previous gives errors.
Open the provider settings and turn off "Use Responses API".
The Responses API endpoint (/v1/responses) works against CodeBurst,
but OpenCat's model-browse probe doesn't currently populate against it. Switch
OpenCat to Chat Completions mode and both browse + chat work immediately.
Your API key is missing or wrong. Newly minted keys are 43 random
characters (base64url). Older keys may start with cb_ — both
are accepted. Generate a fresh one at /api-keys.
Your account is out of credits. Add some on the dashboard, or use a
free-tier model like codeburst-free that doesn't consume credits.
The model name you typed isn't recognized. Run the curl test below to see
the list. Custom {username}-* chains only appear once you've
defined them in your account.
curl https://codeburst.ai/v1/chat/completions \
-H "Authorization: Bearer YOUR_KEY_HERE" \
-H "Content-Type: application/json" \
-d '{"model": "codeburst-free", "messages": [{"role":"user","content":"hello"}]}'