One API balance for Core and Pro abliterated models. Use prepaid credit or a subscription with monthly credit.
The gateway and account console are implemented. Both model routes passed small live tool-calling and streaming checks. Public access awaits production payment and email setup; these checks are not coding benchmarks.
Bring your existing client
The preview implements the OpenAI-compatible POST /v1/chat/completions and GET /v1/models endpoints. Set the base URL to your Ablitron deployment followed by /v1, and use a key from the console.
Supported in the preview
- Text messages, streaming responses and function tool calls.
- Core and Pro through the same customer key and balance.
- Usage metering, cached input rates and revocable keys.
- Monthly allowances and prepaid credit, with no automatic top-ups.
Compatibility limits
This is chat-completions compatibility, not the entire OpenAI API. Responses API, files, images, audio, embeddings, batches and multiple completions per request are not supported in this preview. Specific coding tools must be tested against the endpoints they require.
Predictable spending
Each request reserves credit for a conservative input estimate and its output limit. Final billing uses reported input, cached input and output tokens. Reasoning tokens are included in output usage. Unused reserved credit returns to the original balance. See rates and plans.