AI/ML API

AI/ML API hosts models from OpenAI, Anthropic, Google, Meta, and other providers behind an OpenAI-compatible API.

OpenAI Compatibility

Promptfoo uses the OpenAI provider request format. Supported parameters depend on the model and endpoint.

Setup

To use AI/ML API, you need to set the AIML_API_KEY environment variable or specify the apiKey in the provider configuration.

Example of setting the environment variable:

export AIML_API_KEY=your_api_key_here

Get your API key at aimlapi.com.

Provider Formats

Chat Models

aimlapi:chat:<model_name>

Completion Models

aimlapi:completion:<model_name>

Embedding Models

aimlapi:embedding:<model_name>

Shorthand Format

You can omit the type to default to chat mode:

aimlapi:<model_name>

For Claude 5, set omitDefaults: true to omit Promptfoo's default temperature: 0. Leave sampling parameters unset in your config and environment; explicit values still apply.

Configuration

Configure the provider in your promptfoo configuration file:

promptfooconfig.yaml

# yaml-language-server: $schema=https://promptfoo.dev/config-schema.json
providers:
  - id: aimlapi:chat:google/gemini-3-5-flash
    config:
      temperature: 0.7
      max_tokens: 2000

Configuration Options

Common OpenAI parameters include:

Parameter	Description
apiKey	Your AI/ML API key
temperature	Controls randomness (0.0 to 2.0)
max_tokens	Maximum number of tokens to generate
top_p	Nucleus sampling parameter
frequency_penalty	Penalizes frequent tokens
presence_penalty	Penalizes new tokens based on presence
stop	Sequences where the API will stop generating

Promptfoo requests complete responses; this provider does not support streaming.

Popular Models

Use the model ID shown in the AI/ML API catalog, including its publisher prefix. Examples:

Reasoning Models

• GPT-5.6 Luna: openai/gpt-5.6-luna

Advanced Language Models

• Claude Sonnet 5: anthropic/claude-sonnet-5
• Gemini 3.5 Flash: google/gemini-3-5-flash

Open Source Models

• DeepSeek V4 Pro: deepseek/deepseek-v4-pro

Embedding Models

• Text Embedding 3 Large: openai/text-embedding-3-large

Example Configurations

Basic Example

promptfooconfig.yaml

# yaml-language-server: $schema=https://promptfoo.dev/config-schema.json
providers:
  - aimlapi:chat:deepseek/deepseek-v4-pro
  - aimlapi:chat:openai/gpt-5.6-luna
  - id: aimlapi:chat:anthropic/claude-sonnet-5
    config:
      omitDefaults: true

prompts:
  - 'Explain {{concept}} in simple terms'

tests:
  - vars:
      concept: 'quantum computing'
    assert:
      - type: contains
        value: 'qubit'

Advanced Configuration with Multiple Models

promptfooconfig.yaml

# yaml-language-server: $schema=https://promptfoo.dev/config-schema.json
providers:
  - id: aimlapi:chat:deepseek/deepseek-v4-pro
    label: 'DeepSeek V4 Pro'
    config:
      max_tokens: 4000

  - id: aimlapi:chat:openai/gpt-5.6-luna
    label: 'GPT-5.6 Luna'

  - id: aimlapi:chat:google/gemini-3-5-flash
    label: 'Gemini 3.5 Flash'
    config:
      temperature: 0.5

prompts:
  - 'Write Python code to {{task}}. Return only the code, without Markdown fences.'

tests:
  - vars:
      task: 'implement a binary search tree in Python'
    assert:
      - type: python
        value: |
          # Verify the code is valid Python
          import ast
          try:
            ast.parse(output)
            return True
          except SyntaxError:
            return False
      - type: llm-rubric
        value: 'The code should include insert, search, and delete methods'

Embedding Example

Embedding models back the similar assertion. Set them under defaultTest.options.provider.embedding; an embedding model cannot be used as a top-level eval provider. Extra request fields such as dimensions go under config.passthrough.

promptfooconfig.yaml

# yaml-language-server: $schema=https://promptfoo.dev/config-schema.json
providers:
  - aimlapi:chat:openai/gpt-5.6-luna

defaultTest:
  options:
    provider:
      embedding:
        id: aimlapi:embedding:openai/text-embedding-3-large
        config:
          passthrough:
            dimensions: 1024 # Optional: reduce embedding dimensions

prompts:
  - 'Describe {{topic}} in one sentence.'

tests:
  - vars:
      topic: 'a fox jumping over a dog'
    assert:
      - type: similar
        value: 'The quick brown fox jumps over the lazy dog'
        threshold: 0.7

JSON Mode Example

promptfooconfig.yaml

# yaml-language-server: $schema=https://promptfoo.dev/config-schema.json
providers:
  - id: aimlapi:chat:openai/gpt-5.6-luna
    config:
      response_format: { type: 'json_object' }

prompts:
  - |
    Extract the following information from the text and return as JSON:
    - name
    - age
    - occupation

    Text: {{text}}

tests:
  - vars:
      text: 'John Smith is a 35-year-old software engineer'
    assert:
      - type: is-json
      - type: javascript
        value: |
          const data = JSON.parse(output);
          return data.name === 'John Smith' && 
                 data.age === 35 && 
                 data.occupation === 'software engineer';

Getting Started

Test your setup with working examples:

npx promptfoo@latest init --example provider-aiml-api

Notes

Check AI/ML API pricing for rates and account limits.