feat: add Xiaomi MiMo provider support

[felipebrgs1]

Summary

Add Xiaomi MiMo as a new AI provider with support for MiMo V2 Pro, V2 Omni, and V2 TTS models.

Context

Xiaomi MiMo is a new AI provider offering high-performance models with competitive capabilities. Adding this provider expands the available AI options for users, particularly for those in Asia-Pacific regions with dedicated cluster endpoints.

Changes

• Added XIAOMI_MIMO provider constant to ProviderId enum
• Included Xiaomi MiMo in built-in providers list
• Added display name mapping for "xiaomi_mimo" -> "XiaomiMimo"
• Added parser support for "xiaomi_mimo" string
• Added provider configuration with 3 models:
  • MiMo V2 Pro: 1T+ parameters, 1M context length, supports tools and reasoning
  • MiMo V2 Omni: Multi-modal (text + image), 262K context, supports tools and reasoning
  • MiMo V2 TTS: Text-to-speech model, 8K context
• Configured multi-region API endpoints (Singapore, China, Amsterdam)
• Added unit tests for provider parsing, display name, and built-in list inclusion

Key Implementation Details

• API key configured via XIAOMI_MIMO_API_KEY environment variable
• Dynamic cluster URL selection via CLUSTER parameter (defaults to Singapore)
• Response type compatible with OpenAI format
• Models support parallel tool calls and reasoning where applicable

Use Cases

• Users can now select Xiaomi MiMo models for AI-assisted coding tasks
• Multi-modal support via MiMo V2 Omni for image-based workflows
• Text-to-speech capabilities via MiMo V2 TTS for audio output
• Regional endpoint selection for improved latency in Asia-Pacific

Testing

# Run provider tests
cargo test -p forge_domain provider::tests

# Run all tests to verify no regressions
cargo test

Links

• Xiaomi MiMo documentation: https://xiaomimimo.com

[CLAassistant]

All committers have signed the CLA.

[Copilot]

Pull request overview

Adds Xiaomi MiMo as a new built-in AI provider, enabling users to select MiMo V2 Pro / Omni / TTS models via Forge’s provider abstraction.

Changes:

• Added xiaomi_mimo provider entry to the embedded provider registry (endpoints + 3 hardcoded models).
• Extended ProviderId with XIAOMI_MIMO, including built-in list inclusion, display-name mapping, and FromStr parsing.
• Added unit tests covering parsing, display name, and built-in provider inclusion.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 2 comments.

File	Description
crates/forge_repo/src/provider/provider.json	Registers the xiaomi_mimo provider, its endpoint templating, and hardcoded model metadata.
crates/forge_domain/src/provider.rs	Adds XIAOMI_MIMO to ProviderId (constants, display name, parsing) and unit tests.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

Model mimo-v2-omni is declared with input_modalities of ["text", "image"], but its description claims it natively processes "image, video, and audio". This is misleading for users and inconsistent with the declared capabilities; either update the description to match the supported modalities or extend input_modalities if video/audio are actually supported by the rest of the stack.

[Copilot]

url_param_vars includes CLUSTER, which makes CLUSTER mandatory when creating credentials from environment variables (see provider_repo reading all url_param_vars and erroring if any are missing). That conflicts with the template’s {{else}} token-plan-sgp... default and the PR description (“defaults to Singapore”). Consider making CLUSTER optional (e.g., remove it from url_param_vars and hardcode the default URL), or adjust credential/env handling so missing CLUSTER doesn’t block configuration while still allowing UI selection.

Suggested change

	"url_param_vars": [
	{ "name": "CLUSTER", "options": ["token-plan-sgp.xiaomimimo.com", "token-plan-cn.xiaomimimo.com", "token-plan-ams.xiaomimimo.com"] }
	],
	"url_param_vars": [],