Models

List Text Models (Detailed)

Returns all available text generation and community text models with pricing, capabilities, and metadata including context window size, supported modalities, and tool support. When authenticated: the owner's private community models are included, models are filtered by API key permissions, and paid_only models are hidden if the account has no paid balance or the key only spends Quest Pollen. Each entry includes a health field (status, success_rate, requests) summarizing the last 50 eligible final requests within seven days for community proxies and the last 24 hours for other models, refreshed roughly every 60s. Public community proxy models at or below 80% success are omitted from lists; official models, agents, and models without observations remain listed. Use ?reliability=all to bypass this reliability filter. Pass ?community=false to exclude community models or ?community=true to return only community models.

reliability'reliable' | 'all'

Defaults to reliable: public community proxy models need more than 80% success across the last 50 eligible requests within seven days, or no observations. Official models, agents, and private models are unaffected. Fallback rescues count as successes. Use all to bypass only this discovery filter; permissions and private visibility still apply. Exact-ID calls and fallback routing are unaffected.

source'official' | 'community'

Filter by source. Omit for both official and community models.

community'0' | '1' | 'true' | 'false'

Legacy source filter: true/1 for community, false/0 for official.

querystring

Search the canonical name, aliases, title, description and publisher. Case-insensitive; every whitespace-separated word must match.

capabilitiesstring

Comma-separated capabilities (tool_calling, reasoning, web_search, code_execution, pollinations_models). A model must have all of them.

agent'0' | '1' | 'true' | 'false'

true/1 returns only agents, false/0 excludes agents. Omit for both.

limitinteger

Return at most this many models (1-500), after every other filter and in catalog order.

pollinations-model-source'official' | 'community'

Filter by source. Omit for both official and community models.

pollinations-model-reliability'reliable' | 'all'

Defaults to reliable: public community proxy models need more than 80% success across the last 50 eligible requests within seven days, or no observations. Official models, agents, and private models are unaffected. Fallback rescues count as successes. Use all to bypass only this discovery filter; permissions and private visibility still apply. Exact-ID calls and fallback routing are unaffected.

Response

Success

namestring required

aliasesstring[] required

publisherstring required

Human-readable model publisher, not the inference provider

brand_urlstring uri

brand_icon_urlstring

communityboolean required

agentboolean

base_modelstring

per_user_rpmnumber nullable

pricing_default_labelstring

resolutionsstring[]

titlestring required

descriptionstring

input_modalitiesstring[]

output_modalitiesstring[]

required_safetystring[]

supported_endpointsstring[]

video_capabilitiesstring[]

min_durationnumber

max_durationnumber

default_durationnumber

allowed_durationsnumber[]

duration_stepnumber

max_reference_imagesinteger

max_reference_videosinteger

capabilitiesstring[] required

supported_parametersstring[]

Controls honored by this model through /v1/chat/completions; omitted when unverified and not applicable to /v1/responses.

toolsboolean

supports_structured_outputboolean

Whether JSON and JSON-schema output are supported; omitted when unverified.

max_completion_tokensinteger

Maximum requested output tokens, including max_tokens, max_completion_tokens and max_output_tokens.

reasoningboolean

supports_streamingboolean

False when the upstream does not stream. stream: true still returns a valid event stream, but the whole answer arrives at once when generation finishes. Omitted when the model streams.

context_lengthnumber

voicesstring[]

is_specializedboolean

paid_onlyboolean required

alphaboolean

flat_rateboolean

pricing_unitsobject

Billed unit for pricing fields whose name does not describe it, keyed by pricing field. Rate per unit = pricing[field] × quantity, with quantity defaulting to 1. Example: qwen/qwen-image-2.1 bills completionImageTokens in millionths of a megapixel, so pricing.completionImageTokens 0.00000002 with { unit: megapixel, quantity: 1000000 } is 0.02 Pollen per megapixel.

added_datenumber

List Text Models (Detailed)

Request

Query parameters

Headers

Response

Changes