Models and providers

The console has two pages for finding what to call: Models lists every model on the network, and Providers lists every provider (seller) your gateway can route to. Both open an item in a panel on the right with copyable model ids and a ready-to-run API call. Every member can open both pages.

The gateway endpoint

The top bar of every console page shows the gateway's API base URL with a copy button: your public URL plus /v1 (for example https://gateway.example.com/v1), or the address you opened the console on. Point your clients and SDKs at it. See Calling the API.

Automatic and pinned routing

A request names its model in one of two ways:

Model id	Routing
deepseek-v4-flash	Automatic. The gateway picks the cheapest provider that passes your minimum trust score and routing policy; trust only breaks ties.
<peerId>@deepseek-v4-flash	Pinned. Every request goes to that provider. The peer id is lowercase, without 0x.

Routing policies apply to pinned requests too: a pinned provider that your policy blocks is refused.

Models page

Models lists every model offered on the network, with its type, number of providers, input and output price per million tokens, and context length. Filter by name or alias, by type, or turn on Free to see only models with free offers. A TEE badge means at least one provider runs the model in a trusted execution environment.

Click a model to open its panel:

• The model id with a copy button, its type, context length, capabilities (tool use, reasoning, structured output, vision) and aliases.
• Providers: every provider offering the model, with its price, trust score and latency. Each row has a copy button for the pinned id <peerId>@<model>, verified domain or GitHub links when the provider has them, and a link to the provider on Providers.
• API call: a request for the model. The selector defaults to Best provider (automatic). Pick a provider in it, or click Pin on a provider row, to get the pinned version instead.

Image models get a /v1/images/generations call. Video models are listed but are not served through the gateway's /v1 API.

The Models page shows the whole network, without any routing policy applied. A client calling GET /v1/models with an API key that has an allow list of models only sees the models that key may use.

Providers page

Providers lists the providers the gateway's buyer knows, with prices, trust score, reputation (when available), the latency this gateway measured, and how many requests this gateway sent them in the last 24 hours. Filter by name, peer id or model, or turn on Domain or GitHub, TEE or Free. Badges show TEE, Free, Flagged and Cooling down (a provider that failed recently and is skipped for a while).

Click a provider to open its panel:

• Its peer id with a copy button, verified links, trust score, stake, network usage share, latency and recent failures.
• Allow, Prefer and Block, which add this provider to a routing policy. You choose the scope in Apply to: the gateway default, a workspace's organization or workspace policy, a member or an API key. The other settings of that policy stay as they are. See Routing policies and presets.
• Models: each model the provider offers, with its price, the API it speaks, a copy button for <peerId>@<model> and an API call button.
• API call: ready-to-run requests pinned to this provider.

Below the list, Peer lists holds named groups of providers that routing policies can allow or block in one step.

Panel addresses can be bookmarked or shared: /console/models/<model> and /console/providers/<peerId>. Older /console/network/… links open Providers.

Verified identity

Some providers prove they control a domain or a GitHub account. The console shows those as links on the provider's row and panel. It shows only claims the buyer verified. Providers set this up as described in Verified identity.

API call snippets

Every snippet uses your gateway's base URL and reads the key from $ANTSEED_API_KEY, never a real secret. The format follows the model: chat completions for text models, the images API for image models, and for a pinned provider the API that provider speaks (/v1/chat/completions, /v1/messages or /v1/responses).

curl

Terminal

curl https://gateway.example.com/v1/chat/completions \
  -H "Authorization: Bearer $ANTSEED_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"<peerId>@deepseek-v4-flash","messages":[{"role":"user","content":"Hello"}]}'

Python (OpenAI SDK)

python

import os
from openai import OpenAI

client = OpenAI(base_url="https://gateway.example.com/v1", api_key=os.environ["ANTSEED_API_KEY"])

reply = client.chat.completions.create(
    model="<peerId>@deepseek-v4-flash",
    messages=[{"role": "user", "content": "Hello"}],
)
print(reply.choices[0].message.content)

JavaScript (OpenAI SDK)

JavaScript

import OpenAI from 'openai'

const client = new OpenAI({
  baseURL: 'https://gateway.example.com/v1',
  apiKey: process.env.ANTSEED_API_KEY,
})

const reply = await client.chat.completions.create({
  model: '<peerId>@deepseek-v4-flash',
  messages: [{ role: 'user', content: 'Hello' }],
})
console.log(reply.choices[0].message.content)

Models that speak the Messages API also get an Anthropic SDK tab. On the Providers panel, text models also get Codex, Claude Code and Cursor / OpenAI-compatible tabs; Calling the API shows them.

For the image model qwen-image, the curl tab looks like this:

Terminal

curl https://gateway.example.com/v1/images/generations \
  -H "Authorization: Bearer $ANTSEED_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"<peerId>@qwen-image","prompt":"A lighthouse on a cliff at dawn, watercolor","n":1}'

From the command line

antseed gateway peers lists the providers the running buyer knows, with trust, TEE, prices and this gateway's own 24-hour stats. antseed gateway peers --model deepseek-v4-flash narrows it to one model. See Commands.

Next: Chat, or call the API from your own code.