fix: update Google/Vertex connection test model to gemini-2.5-flash

[pelazas]

Description

Testing a perfectly valid Google AI key on the Manage Models page fails with a 404, because the connection test calls gemini-2.0-flash — which Google has retired. Same for the Vertex entry.

This swaps both TEST_MODELS entries to gemini-2.5-flash, as suggested in the issue thread. Also updated the stale TEST_MODELS snippet in open_notebook/ai/CLAUDE.md so the docs match the code.

Left alone on purpose: the static VERTEX_MODELS discovery list in api/credentials_service.py also carries some older model names, but that's model discovery, not the connection test — happy to do a follow-up if you want it refreshed too.

Related Issue

Fixes #970

Type of Change

• Bug fix (non-breaking change that fixes an issue)
• New feature (non-breaking change that adds functionality)
• Breaking change (fix or feature that would cause existing functionality to not work as expected)
• Documentation update
• Code refactoring (no functional changes)
• Performance improvement
• Test coverage improvement

How Has This Been Tested?

• Tested locally with Docker
• Tested locally with development setup
• Added new unit tests
• Existing tests pass (uv run pytest)
• Manual testing performed (describe below)

Test Details:

Added a regression test in tests/test_credentials_api.py pinning the google/vertex test models to gemini-2.5-flash (it fails on main, passes here). Full suite: 223 passed. ruff check clean. I didn't run a live key test against Google's API, but the change is exactly the model swap suggested by @lfnovo in the issue.

Design Alignment

Which design principles does this PR support? (See DESIGN_PRINCIPLES.md)

• Privacy First
• Simplicity Over Features
• API-First Architecture
• Multi-Provider Flexibility
• Extensibility Through Standards
• Async-First for Performance

Explanation:

Keeps the Google/Vertex provider path actually usable — a key test that always fails defeats the point of supporting the provider.

Checklist

Code Quality

• My code follows PEP 8 style guidelines (Python)
• My code follows TypeScript best practices (Frontend)
• I have added type hints to my code (Python)
• I have added JSDoc comments where appropriate (TypeScript)
• I have performed a self-review of my code
• I have commented my code, particularly in hard-to-understand areas
• My changes generate no new warnings or errors

Testing

• I have added tests that prove my fix is effective or that my feature works
• New and existing unit tests pass locally with my changes
• I ran linting: make ruff or ruff check . --fix
• I ran type checking: make lint or uv run python -m mypy . (mypy isn't in the dev deps of my env; change is a string-literal swap)

Documentation

• I have updated the relevant documentation in /docs (if applicable) — updated the TEST_MODELS snippet in open_notebook/ai/CLAUDE.md
• I have added/updated docstrings for new/modified functions
• I have updated the API documentation (if API changes were made)
• I have added comments to complex logic

Database Changes

• I have created migration scripts for any database schema changes (in /migrations)
• Migration includes both up and down scripts
• Migration has been tested locally

Breaking Changes

• This PR includes breaking changes
• I have documented the migration path for users
• I have updated MIGRATION.md (if applicable)

Screenshots (if applicable)

N/A

Additional Context

The issue is labeled ready + good first issue and had been quiet for a bit, so I picked it up — hope that's fine. Happy to adjust if you'd rather use a different model name.

Pre-Submission Verification

• I have read CONTRIBUTING.md
• I have read DESIGN_PRINCIPLES.md
• This PR addresses an approved issue that was assigned to me (approved/ready, but not formally assigned — see Additional Context)
• I have not included unrelated changes in this PR
• My PR title follows conventional commits format (e.g., "feat: add user authentication")

[cubic-dev-ai]

No issues found across 3 files

Confidence score: 5/5

• Automated review surfaced no issues in the provided summaries.
• No files require special attention.

_{Re-trigger cubic}

[lfnovo]

Closing as a duplicate — the broader cleanup in #1027 (just merged) subsumes this, and it credits your PR. Thank you for catching the root issue (#970) first and for the regression test; I ported exactly that test into #1027, pinned to gemini-3.5-flash rather than gemini-2.5-flash. Heads-up for context: gemini-2.5-flash (which this PR used) is already scheduled for shutdown on 2026-10-16, so it would have re-broken the connection test in ~3 months — gemini-3.5-flash is the current stable GA and Google's named replacement. Appreciate the fix!

[lfnovo]

Follow-up for context: #1035 addresses the root cause you first flagged in #970. Rather than tracking Google's model ids (each has a hard-shutdown date, so it's a treadmill), it switches the Google/Vertex test to Google's floating gemini-flash-latest alias and reframes the connection test so a retired model reports the key as valid, not the connection as broken. Your regression test lives on (updated to pin the alias). Thanks for surfacing this first.