Models

Every model the API serves, by id. Prices are in US dollars and the same on a plan. These tables were updated on 2026-11-11; the live source of truth is GET /v1/models for chat and GET /v1/usage/prices for every model and price.

Chat models

Model id	Name	Price per 1M tokens (input / output)	Notes
`deepseek-v4.1-flash-uncensored`	DeepSeek V4.1 Flash Unfiltered	$0.28 / $1.00	The default. Uncensored
`deepseek-v4.1-flash`	DeepSeek V4.1 Flash	$0.30 / $1.00
`glm-5.3-flash`	GLM-5.3 Flash	$0.15 / $0.50	The standard model
`glm-5.3-flash-uncensored`	GLM-5.3 Flash Unfiltered	$0.25 / $1.00	The same model, uncensored

Cached input is cheaper still: see Chat prices. All four write up to 32,768 output tokens per request. The same ids work on Chat Completions, Messages and Responses.

Uncensored models and routing

When you ask for an -uncensored model, a request that plainly does not need it may be answered by its standard sibling, which is faster and cheaper. On Chat Completions, two response headers tell you what happened:

x-aderra-model-used: the model that answered. You pay that model’s price.
x-aderra-route: why, for example kept:needs_unfiltered or downgraded-from:glm-5.3-flash-uncensored.

Thinking

GLM-5.3 Flash thinks before it answers in Chat Completions, and thinking counts as output tokens. If you set max_tokens below 16,384 and send no reasoning_effort, Aderra sends "reasoning_effort": "minimal" for you, so the budget goes to the answer. Send your own reasoning_effort to choose. DeepSeek V4.1 Flash answers without thinking by default, and setting reasoning_effort on it turns thinking on. Messages and Responses answer without extended thinking unless you ask for it.

List models

GEThttps://api.aderra.ai/v1/models

Returns the chat models your key can call, in the OpenAI list format. Authenticate with Authorization: Bearer or x-api-key.

Each entry has input_modalities, an Aderra addition that OpenAI clients ignore. A model whose list lacks image cannot read images: an image sent to it is not seen. Check it before you attach images in Chat completions.

The body also carries a models array: the same models in the catalogue format Codex reads. Other clients can ignore it.

curl https://api.aderra.ai/v1/models \
  -H "Authorization: Bearer $ADERRA_API_KEY"

from openai import OpenAI
import os

client = OpenAI(base_url="https://api.aderra.ai/v1",
                api_key=os.environ["ADERRA_API_KEY"])

for model in client.models.list():
    print(model.id)

import OpenAI from 'openai';

const client = new OpenAI({
  baseURL: 'https://api.aderra.ai/v1',
  apiKey: process.env.ADERRA_API_KEY,
});

for await (const model of client.models.list()) {
  console.log(model.id);
}

{
  "object": "list",
  "data": [
    {
      "id": "glm-5.3-flash",
      "object": "model",
      "created": 0,
      "owned_by": "aderra",
      "input_modalities": ["text"]
    }
  ],
  "models": []
}

Response: application/json

Image models

For /v1/images/generations and /v1/images/edits. Priced per image. Filtered models apply their own content filter; unfiltered ones don’t (explicit content is allowed, within the content policy).

Model id	Name	Content	Sizes (first is the default)	Price per image	Notes
`gpt-image-2.6`	GPT Image 2.6	Filtered	`1024x1024`, `1536x1024`, `1024x1536`	About $0.12 for a plain image, at most $0.30	Best quality, and the best at text in images. Billed by what each image actually takes to make
`seedream-v45`	Seedream 4.5	Unfiltered	`2048x2048`, `2560x1440`, `1440x2560`	$0.08
`nano-banana-2`	Nano Banana 2	Filtered	`1024x1024`	$0.16
`z-image-turbo`	Z-Image Turbo	Unfiltered	`1024x1024`	$0.024
`flux-schnell`	Flux Schnell	Unfiltered	`1024x1024`	$0.001	The default when `model` is left out. Fast drafts
`gpt-image-2.5`	GPT Image 2.5	Filtered	`1024x1024`, `1536x1024`, `1024x1536`	$0.30	Legacy: use `gpt-image-2.6` instead

Each response’s billed_micro gives the exact price of that image.

Video models

For /v1/videos/generations. Clips run 5 to 15 seconds, priced per second of video. The price depends on the mode as well as the model.

Model id	Name	Modes and price per second	Notes
`minimax-h3`	MiniMax H3	`text` $0.08, `image` $0.08, `reference` $0.18	The default for `text`, `image` and `reference`. 768p
`wan-edit`	Wan 3.0 Video Edit	`edit` $0.20	The default for `edit`. 720p
`minimax-h3-spicy`	MiniMax H3 Spicy	`image` $0.08	Uncensored, within the content policy: `image` with `explicit`
`sd-fast`	Seedance 2.0 Fast	`text` $0.24, `reference` $0.30	Uncensored: `text` and `reference` with `explicit`. References are earlier clips only, not images. 720p
`sd-edit`	Seedance 2.0 Fast Video Edit	`edit` $0.26	Uncensored: `edit` with `explicit`. 720p
`minimax-h3-open`	MiniMax H3 Open	`image` $0.10	Open-weights variant at 480p. Accepts some prompts the default refuses, and keeps people as they are in the image

A 5-second clip costs five times the price per second: $0.40 for a text clip on minimax-h3. An uncensored model is used only when the prompt asks for nudity or sex; any other prompt goes to the standard model, at its price.

Speech model

Model id	Voices	Price	Notes
`aderra-voice-1`	200, every one in every language. Listed by `GET /v1/audio/voices`	$0.03 per 1,000 characters	Streamed as it is generated. Languages, formats and limits from `GET /v1/audio/models`. See Text to speech