AI Gateway · Model catalog
Docs / AI Gateway

Model catalog

Browse hosted models across every modality, plus your own GPU model servers, in one unified catalog.

The catalog lists hosted models reachable via BYOK. Each id is provider/model; filter by modality, provider, or attributes like ZDR and free-tier. (Your own GPU servers are a separate feature — see Model servers.)

Base: https://api.cloud.nevtan.com/api/v1
GET/gateway/modelsList the unified catalog. Supports filters below.

Query parameters

ParameterTypeDescription
typestringModality: text, code, image, video, embed, rerank, realtime, speech, transcription, or all.
providerstringFilter by serving provider slug, e.g. anthropic.
zdrbooleanOnly Zero-Data-Retention models.
freeTierbooleanOnly free-tier models.
qstringFree-text search over id and name.
sortstringreleased (default), price, latency, throughput, or name.

Example model ids

Model idModalityProvider
anthropic/claude-sonnet-5TextAnthropic
openai/gpt-4oTextOpenAI
gemini/gemini-3-flashTextGoogle
openai/text-embedding-3-largeEmbedOpenAI
xai/grok-imagine-video-1.5VideoxAI
openai/whisper-large-v3TranscriptionOpenAI

Browse the catalog in the console

The Model List screen is the visual version of this endpoint — the same models, with filters and a sortable table instead of query parameters.

  1. 1
    Open Model List
    In the AI Gateway, open Model List.
  2. 2
    Filter by modality
    Use the tabs (All, Text, Code, Image, Video, Embed, Rerank, Realtime, Speech, Transcription) to narrow to the kind of model you need.
  3. 3
    Narrow by provider and sort
    Pick a provider from the dropdown, and sort by Release Date, Price, Latency, Throughput, or Name.
  4. 4
    Search by name
    Type any part of a model id or name to jump straight to it.

What each column shows

ColumnWhat it shows
ModelThe provider/model id plus the display name.
Input / OutputPrice — per 1M tokens for text models, or per image / per minute / per second for media models.
ContextContext-window length (e.g. 200K).
CapabilitiesTags for what the model can do.
ProvidersWhich upstream providers can serve the model.
ZDRWhether a Zero-Data-Retention route is available.
No TrainingWhether the provider will not train on your data.
Free TierWhether a free tier is available.
ReleasedThe release date.
The Model List page with filters and the model table
Filter and sort the full hosted catalog — every column comes straight from the models endpoint.
Tip
Switch models by changing one string — the request and response shape stay identical across providers.