feat: add sora-2-2025-12-08 snapshot for OpenAI video generation

[Kamilbenkirane]

Summary

Adds the latest Sora 2 snapshot (sora-2-2025-12-08) to the OpenAI video generation provider.

Changes

• Added sora-2-2025-12-08 snapshot model with appropriate parameter constraints:
  • Duration: 4/8/12 seconds
  • Aspect Ratio: 16:9, 9:16
  • Resolution: 720p

Model Details

• Model ID: sora-2-2025-12-08
• Display Name: Sora 2 (December 2025)
• Provider: OpenAI
• Status: Latest snapshot version (December 8, 2025)

Validation

• Model validated in playground
• Snapshot version verified against OpenAI API documentation

[claude]

PR Review: Add sora-2-2025-12-08 snapshot

✅ Summary

This PR adds the latest Sora 2 snapshot model (sora-2-2025-12-08) to the OpenAI video generation provider. The implementation follows the established patterns in the codebase.

---

🔍 Code Quality & Best Practices

Strengths:

• ✅ Consistent with existing model definitions (sora-2 and sora-2-pro)
• ✅ Follows the established pattern for model registration
• ✅ Clean and readable code structure
• ✅ Proper use of type hints and imports
• ✅ Appropriate parameter constraints matching other Sora 2 models

Observations:

• The implementation is straightforward and mirrors the existing sora-2 model definition (lines 9-18)
• Code formatting appears consistent with the codebase style

---

🐛 Potential Issues

Date in Display Name:

• ⚠️ The display name "Sora 2 (December 2025)" includes "2025" but the snapshot date is 2025-12-08. Since this is actually December 8, 2024 (as of the current date being 2025-12-11, this would be a future date if it were 2025), there may be a typo in the year.
• Recommendation: Verify the correct year with OpenAI's documentation. If this is actually a December 2024 snapshot, the display name should be "Sora 2 (December 2024)" and potentially the model ID should be sora-2-2024-12-08.

Model ID vs Display Name Mismatch:

• The model ID is sora-2-2025-12-08 (format: YYYY-MM-DD)
• However, there's no context about whether this follows OpenAI's official naming convention
• Recommendation: Double-check that sora-2-2025-12-08 is the exact model ID from OpenAI's API documentation

---

⚡ Performance Considerations

• ✅ No performance concerns - this is a static model definition
• ✅ The model constraints are appropriate and match the hardware limitations (720p resolution, specific durations)
• ✅ No runtime overhead introduced

---

🔒 Security Concerns

• ✅ No security issues identified
• The change only adds metadata and doesn't introduce any new code paths or vulnerabilities

---

🧪 Test Coverage

Missing Tests:

• ❌ No tests added for the new model
• The existing integration test in packages/video-generation/tests/integration_tests/test_video_generation/test_generate.py only tests sora-2 (line 13)

Recommendations:

1. Add integration test: Consider adding a parameterized test case for sora-2-2025-12-08 to verify it works correctly
2. Unit test: While not strictly necessary for a model definition, a test validating that the model is correctly registered in MODELS list would be helpful
3. Documentation: Consider updating any documentation that lists available models

Example test addition:

@pytest.mark.parametrize(
    ("provider", "model", "parameters"),
    [
        # ... existing tests ...
        (
            Provider.OPENAI,
            "sora-2-2025-12-08",
            {"duration": "4", "aspect_ratio": "16:9", "resolution": "720p"},
        ),
    ],
)

---

📝 Additional Observations

Duplication of Constraints:

• All three Sora 2 models (sora-2, sora-2-pro, sora-2-2025-12-08) share identical base parameter constraints
• Nice-to-have (not blocking): Consider extracting common constraints to a shared constant to follow DRY principles:

SORA_2_BASE_CONSTRAINTS = {
    VideoGenerationParameter.DURATION: Choice(options=["4", "8", "12"]),
    VideoGenerationParameter.ASPECT_RATIO: Choice(options=["16:9", "9:16"]),
    VideoGenerationParameter.RESOLUTION: Choice(options=["720p"]),
}

Model Lifecycle:

• Since this is a snapshot model with a specific date, consider:
  • Will this model eventually be deprecated?
  • Should there be documentation about when to use snapshots vs. the base sora-2 model?

---

✅ Verdict

Overall Assessment: Approve with minor recommendations

The code is well-structured and follows existing patterns. The main concerns are:

1. Verify the date/year in both the model ID and display name (potential typo)
2. Add test coverage for the new model
3. Consider adding documentation about model snapshots

The PR is functionally correct and safe to merge after addressing the date verification issue.

---

Review generated with attention to code quality, security, performance, and testing practices 🤖