OpenAI-compatible APIs
Use a familiar API shape for chat completions, responses, embeddings and common SDK integrations.
AI Model Hub for OpenAI, Claude, Gemini and Multi-Model Routing
TokenOpen gives teams one LLM API gateway to access OpenAI-compatible models, Claude, Gemini, DeepSeek, Qwen, Kimi, Doubao, Hunyuan, Azure OpenAI, AWS Bedrock and other AI model providers while keeping API keys, token usage and billing under one control plane.
Supported model families and protocols
Claude Messages
Connect Claude model traffic through the same gateway, quota, logging and billing controls used by other providers.
Gemini
Unify Gemini access with shared API key management, usage analytics and routing policies.
DeepSeek and Qwen
Route reasoning, coding, long-context and open-weight model traffic based on cost, latency, capability and availability.
Kimi and Moonshot
Manage Kimi and Moonshot model access alongside the same API key, quota and usage reporting workflow.
Doubao and VolcEngine
Connect Doubao chat, vision, image and video model tasks where channels are configured for production workloads.
Tencent Hunyuan
Bring Hunyuan model calls into centralized gateway routing, token accounting and team access governance.
Azure OpenAI
Bring enterprise cloud model deployments into the same token accounting and access management workflow.
AWS Bedrock
Manage Bedrock model channels alongside other upstream providers for resilient production AI workloads.
Why a model hub belongs inside the token platform
Multi-model AI adoption creates operational work beyond choosing a model. Teams need API key isolation, model permissions, token analytics, quota control, fallback routing and cost attribution across every provider. TokenOpen combines the model hub, LLM API gateway and AI token platform into one management layer.
| Need | Model-only directory | TokenOpen Model Hub |
|---|---|---|
| Provider access | Lists available models or providers. | Connects international and domestic providers through one API gateway and shared authentication layer. |
| Token visibility | Usually leaves usage tracking to each provider. | Tracks prompt, completion, cached and billed tokens by model, channel, user and key. |
| Production routing | Does not control live request routing. | Supports multi-channel routing, fallback and operational visibility for AI traffic. |
| Team governance | Limited permission and quota management. | Uses API keys, groups, quota rules and billing policies to govern model access. |
Common use cases
AI application teams
Switch or compare models without rewriting authentication, SDK setup and usage accounting.
SaaS platforms
Allocate model quotas and API keys to customers, workspaces or internal departments.
Operations and finance
Understand token cost by provider, model and business unit before spend becomes difficult to control.
Model Hub FAQ
Which AI model providers does TokenOpen Model Hub support?
TokenOpen supports unified access to OpenAI-compatible models, Claude, Gemini, DeepSeek, Qwen, Kimi, Doubao, Hunyuan, Azure OpenAI, AWS Bedrock and other configured providers through one gateway.
Can TokenOpen manage multimodal, image and video model tasks?
When the corresponding upstream channels are configured, TokenOpen can manage text, vision, image generation, image editing, text-to-video and image-to-video tasks with shared logging, routing and billing controls.
How does TokenOpen track token usage and billing?
TokenOpen records prompt, completion, cached and billed tokens by model, channel, user and API key, so teams can attribute cost and enforce quotas from one control plane.
How does TokenOpen handle model routing and fallback?
TokenOpen lets teams configure multiple upstream channels, model mappings, quota rules and fallback routing so production traffic can continue when a provider is slow or unavailable.