What does this PR do?

Preserves explicit DeepSeek V4 public API model IDs during native DeepSeek model normalization.

Before this change, configuring deepseek-v4-pro or deepseek-v4-flash for the native deepseek provider silently normalized the model to deepseek-chat. That made Hermes appear to work while calling a different model than the user explicitly requested.

This keeps the change narrow: only the DeepSeek accepted-model allowlist is expanded, while legacy aliases and reasoner-style alias normalization keep their existing behavior.

Related Issue

No issue filed.

Type of Change

• 🐛 Bug fix (non-breaking change that fixes an issue)
• ✨ New feature (non-breaking change that adds functionality)
• 🔒 Security fix
• 📝 Documentation update
• ✅ Tests (adding or improving test coverage)
• ♻️ Refactor (no behavior change)
• 🎯 New skill (bundled or hub)

Changes Made

• hermes_cli/model_normalize.py
  • Preserve deepseek-v4-flash and deepseek-v4-pro for the native deepseek provider.
  • Keep deepseek-chat and deepseek-reasoner aliases working.
  • Keep reasoner-style aliases such as deepseek-r1, deepseek-reasoning, and deepseek-think mapping to deepseek-reasoner.
  • Update the DeepSeek normalization comments/docstring to reflect the V4 public model IDs.
• tests/hermes_cli/test_model_normalize.py
  • Add regression tests for bare and deepseek/-prefixed V4 model IDs.
  • Add regression coverage that reasoner-style aliases still map to deepseek-reasoner.

How to Test

1. Run the focused regression tests:

   scripts/run_tests.sh tests/hermes_cli/test_model_normalize.py::TestDeepSeekV4ModelPreservation -v --tb=short

2. Run the full model-normalization test file:

   scripts/run_tests.sh tests/hermes_cli/test_model_normalize.py -v --tb=short

3. Confirm expected behavior, for example:

   normalize_model_for_provider("deepseek-v4-pro", "deepseek") == "deepseek-v4-pro"
   normalize_model_for_provider("deepseek/deepseek-v4-flash", "deepseek") == "deepseek-v4-flash"
   normalize_model_for_provider("deepseek-r1", "deepseek") == "deepseek-reasoner"

Checklist

Code

• I've read the Contributing Guide
• My commit messages follow Conventional Commits (fix(scope):, feat(scope):, etc.)
• I searched for existing PRs to make sure this isn't a duplicate
• My PR contains only changes related to this fix/feature (no unrelated commits)
• I've run pytest tests/ -q and all tests pass
• I've added tests for my changes (required for bug fixes, strongly encouraged for features)
• I've tested on my platform: macOS / Darwin

Documentation & Housekeeping

• I've updated relevant documentation (README, docs/, docstrings) — or N/A
• I've updated cli-config.yaml.example if I added/changed config keys — or N/A
• I've updated CONTRIBUTING.md or AGENTS.md if I changed architecture or workflows — or N/A
• I've considered cross-platform impact (Windows, macOS) per the compatibility guide — or N/A
• I've updated tool descriptions/schemas if I changed tool behavior — or N/A

Screenshots / Logs

Focused regression test:

scripts/run_tests.sh tests/hermes_cli/test_model_normalize.py::TestDeepSeekV4ModelPreservation -v --tb=short
10 passed in 0.37s

Full model-normalization test file:

scripts/run_tests.sh tests/hermes_cli/test_model_normalize.py -v --tb=short
65 passed in 0.41s

Independent local review was also requested from Claude Code Opus 4.7, Kimi 2.6 via Kimi Code CLI, and DeepSeek V4 Pro. All approved with no blocking issues.