---
title: "Get Multimodal Chat Response"
method: POST
path: "/chatbot/v3/response"
tags: ["chatbot"]
---

# Get Multimodal Chat Response

`POST /chatbot/v3/response`

Multimodal chat response endpoint.

It can be used to support engaging conversations with both text and image responses.

For usage examples, see: https://docs.exh.ai/recipes/multimodal-chat

## Headers

- `X-Request-Id` string

## Request body

- ChatbotV3PayloadSchema
  - `context` MessageWithImagePrompt[] — The context of the conversation. Set to empty list to get bot's greeting. Maximum 40 messages.
    - `turn` string, required — The turn of the message (`bot` or `user`)
    - `message` string, required
    - `image_prompt` string, nullable — The prompt for the image that was generated on this turn.
  - `bot_profile` CharacterProfile, required
    - `id` string, nullable — Unique identifier. Provide user's id and bot's id to enable Long-Term Memory - recall of conversation beyond 30 messages
    - `name` string, required
    - `description` string, nullable — Detailed description of your character. It can include traits, history, mannerisms, backstory, things they want to talk about, etc.
    - `appearance` string, nullable — Description of how your character looks like. Some ideas to specify in this field: gender,ethnicity, hair color, body type (slim, athletic, curvy, etc.), age, tattoos, and eye color. This field will affect the generated image and image prompt the most.
    - `pronoun` 'he/him' | 'she/her' | 'they/them'
    - `example_messages` string[], nullable — Example messages to help the model understand the character's personality and style. The model will use these examples to generate a response that is more likely to be similar to the character's style.
  - `user_profile` UserProfile
    - `id` string, nullable — Unique identifier. Provide user's id and bot's id to enable Long-Term Memory
    - `name` string, nullable
    - `description` string, nullable — Detailed description of your character. It can include traits, history, mannerisms, backstory, things they want to talk about, etc.
    - `appearance` string, nullable — Description of how your character looks like. Some ideas to specify in this field: gender,ethnicity, hair color, body type (slim, athletic, curvy, etc.), age, tattoos, and eye color.
    - `pronoun` 'he/him' | 'she/her' | 'they/them'
  - `chat_settings` ChatV3Settings
    - `model_name` 'roleplay' | 'realistic_chat'
    - `allow_nsfw` boolean — For enabling NSFW bot responses support, set the flag to 'True'; otherwise, set it to 'False'.
    - `tasks` string[], nullable — Modifications for LLM instruction. Maximum 5 tasks. Total length of all tasks must not exceed 2500 characters.
    - `enable_memory` boolean — A flag that determines whether the long-term memory feature is used. If it's set to True or not provided, and both bot_profile.id and user_profile.id are also provided, the service will store and utilize memory of the conversation. If it's set to False, no memory of past events beyond the current context will be available. Memory starts accumulating after 20 messages in a chat.
  - `image_settings` ChatV3ImageSettings
    - `identity_image_url` string, uri, nullable — Url to the image with bot's identity picture.
    - `model_name` 'persona-2' | 'persona-3' — Model to use for image generation.
    - `style` 'realistic' | 'anime' — Artistic style for generating the gallery image. If not specified, the style is automatically detected based on the identity image.
    - `allow_nsfw` boolean — For enabling NSFW image support, set the flag to 'True'; otherwise, set it to 'False'.
    - `usage_mode` 'force' | 'on_user_request' | 'off' — The image usage mode. Determines how images are handled in the chat. If set to 'off', images will not be used. If set to 'on_user_request', images will be provided only when requested by the user. If set to 'force', images will always be provided.
    - `return_bs64` boolean — If return_bs64 is set to True, a base64-endcoded generated image and the image prompt will be provided. If set to False, only the text prompt describing the image will be returned.
  - `catalog` ContentItem[], nullable — Content catalog for selling. When provided, the bot will be able to recommend items from this catalog during conversation. Each item should have a name, description, price, and link.
    - `name` string, required
    - `description` string, required
    - `price` number, required
    - `link` string, required

## Response `200`

Successful Response

- ChatbotV3ResponseSchema
  - `response` string, required
  - `image_response` ImageV3ResponseSchema
    - `prompt` string, required
    - `bs64` string, nullable

## Other responses

- `422` — Validation Error

---

[API](https://skmtc.dev/exh/apis/ex-human.md) · [All operations](https://skmtc.dev/exh/apis/ex-human/llms.txt) · [OpenAPI document](https://skmtc-service-production.skmtc.workers.dev/v1/apis/exh/ex-human/revisions/cd3ee850caf2/schema)