self-operating-computer icon indicating copy to clipboard operation
self-operating-computer copied to clipboard

Adds support for GPT-5 models

Open ustincameron opened this issue 4 months ago • 0 comments

What does this PR do?

  • Introduced handling for new models (gpt-5-with-ocr, gpt-5-research-with-ocr, gpt-5-mini-with-ocr, and gpt-5-nano-with-ocr) in apis.py.
  • Updated config.py to include the new models in validation.
  • Changed default model in main.py to gpt-5-with-ocr.
  • Updated requirements.txt for greater flexibility with dependency versions.
  • Bumped the version in setup.py to 1.6.0.
  • Updates OpenAI client dependency to support GPT-5 API endpoints
Screenshot 2025-08-09 at 1 23 07 PM

Requirement/Documentation

  • :white_check_mark: All GPT-5 variants tested and functional
  • :white_check_mark: Regression testing confirms existing models still work
  • :white_check_mark: Default model properly set to GPT-5
  • :white_check_mark: API calls verified with correct OpenAI endpoints
  • :white_check_mark: No breaking changes to existing functionality

Usage Examples

# Use new default (GPT-5)
operate --prompt "Go to [Google.com](http://google.com/)"
# Specific GPT-5 variants
operate -m gpt-5-mini-with-ocr --prompt "Open calculator"
operate -m gpt-5-nano-with-ocr --prompt "Take screenshot"
# Existing models still work
operate -m gpt-4-with-ocr --prompt "Go to GitHub"

Type of change

  • [x] New feature (non-breaking change which adds functionality)
  • [x] Tests (Unit/Integration/E2E or any other test)

Mandatory Tasks

  • [x] Make sure you have self-reviewed the code. A decent size PR without self-review might be rejected. Make sure before submmiting this PR you run tests with evaluate.py

ustincameron avatar Aug 09 '25 20:08 ustincameron