Platform features

OpenAI-compatible access

Yomi API provides OpenAI-compatible access to leading models worldwide through a single interface. Migrate an existing project by replacing base_url and the key. To switch models, change the model parameter without adapting a separate SDK for each provider.

Supported model providers

• OpenAI: GPT-5.4, GPT-5.5 and GPT-5.6 series (sol/terra)
• Anthropic: Claude Opus 4.8, Claude Sonnet 5
• Zhipu: GLM-5.2
• DeepSeek: DeepSeek V4 Pro, DeepSeek V4 Flash
• Moonshot AI: Kimi K3

See "Models" (the Model Catalog) in the console for the complete catalog, route prices and performance data.

Feature coverage

Supported capabilities

• Chat completions: Chat Completions endpoint with streaming and function calling
• Responses endpoint: OpenAI Responses API
• Native Claude format: /v1/messages endpoint
• Image generation and editing: Endpoints are available (/v1/images/generations, /v1/images/edits); see the Model Catalog for available models
• Video generation: Asynchronous endpoints are available (/v1/videos, /v1/video/generations); see the Model Catalog for available models
• Audio processing: /v1/audio/* endpoints are available; see the Model Catalog for available models
• Embeddings: The /v1/embeddings endpoint is available; see the Model Catalog for available models
• Moderation and reranking: /v1/moderations, /v1/rerank
• Realtime endpoint: /v1/realtime (WebSocket)
• Common parameters: temperature, top_p, max_tokens and other OpenAI parameters

Not currently supported

• Fine-tuning API
• Files management API
• Organization management API
• Billing management API

Keys and connection examples

1. Sign in to the console and open "API Keys" in the top navigation.
2. Click Create, enter a name, select a model group and submit.
3. Copy the generated key for your API calls.

The API Keys page also includes configuration examples for Claude Code, Codex CLI and other tools. Any OpenAI-compatible client can use the generic configuration in the "Other" tab.