Skip to main content

QuiverAI

QuiverAI provides an OpenAI-compatible chat API optimized for generating SVG vector graphics. The provider extends the OpenAI provider and supports all standard chat completion options.

Setup​

  1. Get an API key from QuiverAI
  2. Set QUIVERAI_API_KEY environment variable or specify apiKey in your config

Available Models​

ModelDescription
arrow-0.5Flagship SVG generation model
arrow-0.5-previewPreview version with experimental features

Both models support text and image inputs, and output SVG graphics in various styles (flat, outline, duotone, gradient) and modes (icon, illustration, logo).

Provider Formats​

FormatDescription
quiverai:model-nameChat completions (default)
quiverai:chat:model-nameChat completions (explicit)

Configuration​

providers:
  - id: quiverai:arrow-0.5
    config:
      temperature: 0.7
      max_tokens: 4096

All OpenAI configuration options are supported, including temperature, top_p, max_tokens, frequency_penalty, and presence_penalty.

SVG Generation Example​

QuiverAI's Arrow model generates SVG graphics via chat completions:

providers:
  - id: quiverai:arrow-0.5
    config:
      max_tokens: 4096

prompts:
  - file://prompt.json

tests:
  - vars:
      subject: a red heart icon
    assert:
      - type: contains
        value: '<svg'
      - type: contains
        value: '</svg>'

Where prompt.json contains:

[
  {
    "role": "system",
    "content": "You are a vector graphics designer. Output valid SVG code only."
  },
  { "role": "user", "content": "Create an SVG of: {{subject}}" }
]

Sample Outputs​

Here are examples of SVGs generated by QuiverAI's Arrow model:

red heart icon
yellow star
rocket with flames

Environment Variables​

VariableDescription
QUIVERAI_API_KEYAPI key (required)

See Also​