Platform API
Model Management API
Model management API: the endpoint returns model metadata such as pricing, context length, endpoints, and lifecycle, with request examples and filters.
GET
Model Management API
New API Version
Get Model Information
Endpoint:GET https://aihubmix.com/api/v1/models
Description: Fetches detailed information for all available models.
Model Object Field Descriptions
Basic information
The unique identifier for the model, which is the value used for the
model parameter when calling the APIThe platform-internal ID of the model developer, usable with the
developer_id request parameter for filteringModel vendor identifier (such as
openai, anthropic, zhipu). This is a separate numbering system from developer_id. Optional field; when absent, the value has not been recordedModel type, comma-separated. Supported values:
llm (large language model), image_generation (image generation model), video (video generation model), tts (text-to-speech model), stt (speech-to-text model), embedding (embedding model), rerank (reranking model), ocr (text recognition model), search (search model), 3d (3D generation model)Supported features, comma-separated. Supported values:
thinking (supports reasoning), tools (supports tool use), function_calling (supports function calling), web (supports search), deepsearch (supports deep search), long_context (long-context model), structured_outputs (structured outputs)Supported output modalities, comma-separated. Supported values:
text, image, audio, video. Optional field; when absent, the value has not been annotatedAPI protocols available for the model, comma-separated. Supported values:
chat_completions (Chat Completions API), responses (Responses API), claude_api (Claude Messages API), gemini_api (Gemini API). An empty value means the field has not been annotated yetModel release date, in
YYYY-MM-DD or YYYY-MM format. Optional field; when absent, the value has not been recordedDate of the model’s most recent update. When there is no in-place upgrade record, it equals
release_date. Optional fieldWhether the model weights are openly downloadable and self-deployable.
false means the vendor has explicitly stated the weights will not be released. Optional field; when absent, the value has not been recordedPricing
Tiered pricing array, optional field. Each element contains a
tier object (type is always context, size is the context token count threshold) plus the input, output, cache_read, and cache_write prices for that tier. When the request’s context length exceeds size, billing follows that tier’s prices; the top-level input, output, and related fields are the default tier pricesReasoning and tool capabilities
Whether the model has reasoning capability, a boolean normalized from
features. Optional field; when absent, the value has not been annotated, and absence does not mean the capability is unsupportedWhether the model supports tool use, a boolean normalized from
features. Optional field; when absent, the value has not been annotated, and absence does not mean the capability is unsupportedReasoning level configuration array, optional field. Each item contains
type. Supported values: toggle (reasoning can be switched on and off), effort (controlled by level, with values listing the available levels and default giving the default level), budget_tokens (controlled by token budget)Where reasoning content is returned under each API protocol. Optional field; when absent, the value has not been recorded. Keys are protocol names (
anthropic.messages, openai.chat_completions, openai.responses, google.gemini); values are true (the protocol can return reasoning content) or {"field": "reasoning_content"} (reasoning content is returned via that field). Read the key that matches the protocol you actually callVariants and lifecycle
Parent model ID, optional field. When present, the current
model_id and the parent model are the same model, differing only in platform-side tier or presetVariant type, appearing together with
variant_of. Supported values: pricing-tier (pricing tier), channel-tier (access tier), mode-preset (mode preset), deprecated-alias (deprecated alias)Model lifecycle stage, always returned. Supported values:
active (in normal service), deprecated (retirement announced, still callable). Models whose service has ended do not appear in this endpoint’s responsePlanned end-of-service time (ISO 8601 format). Optional field; may be returned only in the
deprecated stageRecommended replacement model ID for migration. Optional field; may be returned only in the
deprecated stageLimited-time promotions
Promotion time rule type. Supported values:
absolute (an absolute time range, see the absolute object’s start and end), daily (fixed time windows each day, see daily.ranges), weekly (fixed time windows each week, see the weekly array’s weekdays and ranges; weekdays uses ISO 8601 numbering, with Monday as 1 and Sunday as 7)Request Examples
Request Parameter Descriptions (for Filtering)
Model type. Supported values:
llm (large language model), image_generation (image generation model), video (video generation model), tts (text-to-speech model), stt (speech-to-text model), embedding (embedding model), rerank (reranking model), ocr (text recognition model), search (search model), 3d (3D generation model)Input modalities. Supported values:
text, image, audio, video, pdf. Supports multi-modality queries (comma-separated)Model features. Supported values:
thinking (supports reasoning), tools (supports tool use), function_calling (supports function calling), web (supports search), deepsearch (supports deep search), long_context (long-context model), structured_outputs (structured outputs). Supports multi-feature queries (comma-separated)Filter by API protocol. Supported values:
chat_completions, responses, claude_api, gemini_api. Supports multi-value queries (comma-separated)When set to
true, returns only models for which the platform has completed request parameter schema verification (models whose schema_checked response field is true)Sort by field. Supported values:
•
•
•
•
•
model_ratio: sort by cost-effectiveness•
context_length: sort by context length•
coding: prioritize coding models•
order: sort by default orderSuccessful Response Example
Usage Scenario Examples
Note: When using smart sort for coding models, the system will prioritize models tagged with coding, and other models will be listed in the default order.
Performance Optimization
Caching Mechanism
- Cache Policy: HTTP caching, cache duration 300 seconds (5 minutes)
- Cache Control:
Cache-Control: private, max-age=300, stale-while-revalidate=300 - Content Validation: Supports ETag content hash validation
Cache Usage Example
If the content has not been updated, the server returns a 304 Not Modified status code.
Error Handling
Important Notes
- Data integrity: This endpoint returns all models that meet the criteria, without pagination
- Type compatibility: Supports automatic mapping between new and old type identifiers
t2t↔llmt2i↔image_generationt2v↔videoreranking↔rerank
- Filtering logic: Multiple filter conditions are combined with a logical AND
- Sorting rule: When no sorting method is specified, models are arranged in the system’s default order
- Lifecycle: Models with
retire_stageset todeprecatedcan still be called normally. Refer to thesuccessor_modelandretired_atfields to plan migration ahead of time; models whose service has ended do not appear in the returned list - Optional fields: When a field marked as optional is absent, the information has not been annotated or recorded. Absence does not mean the model lacks the corresponding capability
Legacy API Version
⚠️ Note: The following are legacy API endpoints. It is recommended to use the new API version for better performance and features.
Get Model List
Endpoint:GET /v1/models
- If a user is logged in, it retrieves the list of available models for the user’s group. If no user is logged in, it retrieves the list for the
defaultgroup. - If the header contains an
Authorizationfield, it queries the list of models configured for the corresponding token.
Response
Response Data Structure
Status Code 200Get Model Information
Endpoint:GET /v1/models/:model
Request Parameters
Response Example:
Response
Response Data Structure
Status Code 200Last updated: 2026-09-18