Current capabilities and future direction
Available in controlled development: deepseek-flash text API. This is the currently implemented model; public production service is not live.
deepseek-flash
Supported subset
- POST /v1/chat/completions with stream=false
- Text-only system / user / assistant messages
- Optional positive integer max_tokens, at most 65,536
- Authenticated GET /v1/models discovery
Not supported: streaming, images, audio, embeddings, tools, Responses API, temperature or top_p. Other models have no announced availability or delivery date.
Planned / Not available yet
Our direction is a multi-provider, multi-model AI API platform, eventually extending beyond text to image, audio and video. These capabilities are not implemented or available yet. No additional provider support, delivery date, pricing or API compatibility is promised.