nirbarazida
/
promptfoo
mirror of https://github.com/promptfoo/promptfoo


  
1

	
2

	
3

	
4

	
5

	
6

	
7

	
8

	
9

	
10

	
11

	
12

	
13

	
14

	
15

	
16

	
17

	
18

	
19

	
20

	
21

	
22

	
23

	
24

	
25

	
26

	
27

	
28

	
29

	
30

	
31

	
32

	
33

	
34

	
35

	
36

	
37

	
38

	
            # yaml-language-server: $schema=https://promptfoo.dev/config-schema.json
description: 'Claude vs GPT Image Analysis'

prompts:
  - ./prompt.py:format_image_prompt
  - ./prompt.js:formatImagePrompt

providers:
  - id: bedrock:anthropic.claude-3-sonnet-20240229-v1:0
  - id: anthropic:claude-3-5-sonnet-20241022
  - id: openai:gpt-4.1
    label: custom label for gpt-4.1

tests:
  - vars:
      image_url: https://upload.wikimedia.org/wikipedia/commons/thumb/0/0d/Great_Wave_off_Kanagawa2.jpg/640px-Great_Wave_off_Kanagawa2.jpg
      label: Great Wave off Kanagawa

defaultTest:
  options:
    provider: openai:gpt-4.1
    rubricPrompt: |
      [
        {
          "role": "system",
          "content": "You are a meticulous grader. Evaluate how accurately the answer describes the provided image. Respond in JSON {reason:string, pass:boolean, score:number}"
        },
        {
          "role": "user",
          "content": [
            {"type": "image_url", "image_url": {"url": "{{image_url}}"}},
            {"type": "text", "text": "Answer: {{ output }}\nCriteria: {{ rubric }}"}
          ]
        }
      ]
  assert:
    - type: llm-rubric
      value: Is a detailed description of the image '{{label}}'