What does this PR do?

Add reasoning_effort (intensity control) and thinking toggle support for the native DeepSeek provider (api.deepseek.com).

DeepSeek V4 models (Flash / Pro) expose two independent thinking controls through their OpenAI-compatible API:

• Thinking toggle: {"thinking": {"type": "enabled/disabled"}} in extra_body
• Effort control: "reasoning_effort": "high" / "max" at top level

The server maps low / medium → high and xhigh → max for compatibility with OpenRouter-style effort levels. This PR mirrors that mapping client-side.

Previously, the deepseek provider profile was a bare ProviderProfile that emitted neither parameter — the native API path had no way to toggle thinking or control reasoning intensity. The OpenRouter path sent a generic extra_body.reasoning block that does not match DeepSeek's native format.

This PR introduces a DeepSeekProfile subclass implementing build_api_kwargs_extras() that emits both parameters, following the same pattern already used by KimiProfile for Kimi / Moonshot.

Related Issue

Fixes #

Type of Change

• ✨ New feature (non-breaking change that adds functionality)
• ✅ Tests (adding or improving test coverage)

Changes Made

• plugins/model-providers/deepseek/__init__.py — Replace bare ProviderProfile with DeepSeekProfile subclass that implements build_api_kwargs_extras(), emitting extra_body.thinking (toggle) and top-level reasoning_effort (intensity). Effort mapping: low/medium/minimal → high, xhigh → max, high → high, max → max, unknown → high.
• tests/providers/test_provider_profiles.py — Add TestDeepSeekProfile class with 12 test cases covering all effort mappings (none, minimal, low, medium, high, xhigh, max, unknown, missing), alias lookup, profile registration, and thinking toggle behavior.

How to Test

1. Run pytest tests/providers/test_provider_profiles.py -v -n 0 — all 53 tests should pass (41 existing + 12 new).
2. Start a Hermes session with provider: deepseek and a DeepSeek V4 model.
3. Use /reasoning none to disable thinking — the API request should include extra_body.thinking = {"type": "disabled"} with no reasoning_effort.
4. Use /reasoning xhigh — the API request should include reasoning_effort: "max" and extra_body.thinking = {"type": "enabled"}.
5. Verify default behavior (no /reasoning set) sends thinking: enabled + reasoning_effort: high.

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: WSL (Ubuntu) / Windows 11

Documentation & Housekeeping

• I've updated relevant documentation (README, docs/, docstrings) — or N/A (docstring on build_api_kwargs_extras updated)
• 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 (provider profile is platform-agnostic)
• I've updated tool descriptions/schemas if I changed tool behavior — or N/A

Screenshots / Logs

tests/providers/test_provider_profiles.py ................................
tests/providers/test_provider_profiles.py::TestDeepSeekProfile::test_profile_registered PASSED
tests/providers/test_provider_profiles.py::TestDeepSeekProfile::test_alias_lookup PASSED
tests/providers/test_provider_profiles.py::TestDeepSeekProfile::test_no_config_defaults PASSED
tests/providers/test_provider_profiles.py::TestDeepSeekProfile::test_thinking_disabled PASSED
tests/providers/test_provider_profiles.py::TestDeepSeekProfile::test_effort_high PASSED
tests/providers/test_provider_profiles.py::TestDeepSeekProfile::test_effort_max PASSED
tests/providers/test_provider_profiles.py::TestDeepSeekProfile::test_effort_xhigh_maps_to_max PASSED
tests/providers/test_provider_profiles.py::TestDeepSeekProfile::test_effort_medium_maps_to_high PASSED
tests/providers/test_provider_profiles.py::TestDeepSeekProfile::test_effort_low_maps_to_high PASSED
tests/providers/test_provider_profiles.py::TestDeepSeekProfile::test_effort_minimal_maps_to_high PASSED
tests/providers/test_provider_profiles.py::TestDeepSeekProfile::test_unknown_effort_falls_back_to_high PASSED
tests/providers/test_provider_profiles.py::TestDeepSeekProfile::test_missing_effort_defaults_to_high PASSED

============================== 53 passed in 1.10s ==============================

Ref: https://api-docs.deepseek.com/guides/thinking_mode