https://api.junji-tech.com/v1API documentation
Junji API currently supports text generation through OpenAI Chat Completions, OpenAI Responses, and Anthropic Messages.
https://api.junji-tech.comAuthentication
OpenAI-compatible clients use a Bearer token. Anthropic clients use X-API-Key. Do not send two different keys in the same request.
Authorization: Bearer jsk_live_your_keyX-API-Key: jsk_live_your_keyEndpoints
/v1/modelsList currently available models/v1/chat/completionsOpenAI Chat Completions/v1/responsesOpenAI Responses/v1/messagesAnthropic MessagesRequest scope
Generation requests use JSON and must include a currently listed model. The first release supports plain-text system, user, and assistant messages plus common sampling controls.
- Maximum request body: 4 MiB.
- Current output limit: 512 tokens.
- Tools, multimodal content, structured output, stateful Responses, and Anthropic thinking are not yet supported. Unsupported input fails explicitly rather than silently degrading.
Streaming
Set stream: true to receive text/event-stream. Each endpoint emits its native SSE event structure; clients should consume the terminal event rather than treating a closed connection as successful completion.
Usage and billing
Accounts use prepaid CNY credit. The active model price is snapped when a request is admitted, then final cost is settled from the actual input and output token counts reported by compute. Usage statistics can take a few seconds to update.