feat: add recent AI models (GPT Image 2, Claude Opus 4.7, Qwen3.6, DeepSeek V4 Pro[1M], Kimi K2.6, Grok 4.20, MiMo V2.5)

[wahajahmed010]

Summary

Refreshes the built-in provider/model registry with recent AI model releases and clarifies server-config extensibility.

Changes

New models added:

• OpenAI: gpt-5.4-pro, gpt-image-2, gpt-image-2-2026-04-21, gpt-image-1.5, gpt-image-1, gpt-image-1-mini, chatgpt-image-latest
• Anthropic: claude-opus-4-7
• Qwen: qwen3.6-max-preview, qwen3.6-plus, qwen3.6-plus-2026-04-02, qwen3.6-flash, qwen3.6-flash-2026-04-16, qwen3.6-35b-a3b
• Qwen Image: qwen-image-2.0-pro, qwen-image-2.0-pro-2026-03-03, qwen-image-2.0, qwen-image-2.0-2026-03-03
• DeepSeek: deepseek-v4-pro[1m] (1M context variant)
• Kimi: kimi-k2.6
• xAI/Grok: grok-4.20-reasoning, grok-4.20, grok-4.20-multi-agent
• Xiaomi MiMo: mimo-v2.5-pro, mimo-v2.5

Documentation:

• Updated .env.example with example model lists for each provider
• Clarified that HY3 is a model name (Tencent Hunyuan), not a provider prefix

Server-config extensibility:

The existing implementation in lib/store/settings.ts already preserves unknown server-provided model IDs from *_MODELS environment variables or YAML config as synthetic { id, name } entries. This means:

• When a server configures QWEN_MODELS=qwen3.7-new-model, it will appear in the UI even if not in the built-in registry
• Newly released models can be used immediately via server config before the app registry is updated
• Aggregator-specific IDs (e.g., OpenRouter routing IDs) stay out of default direct-provider lists unless they are official direct API IDs

Testing

• All model IDs match the official API documentation links provided in the issue
• No breaking changes to existing functionality
• Server-configured models continue to work via the existing info.models.map(id => currentModelMap.get(id) ?? { id, name: id }) pattern

Related

• Addresses [Feature]: Refresh recent AI model support and preserve server-configured model IDs #480
• Related to [Bug]: in provider-keys.md 1.Lowest-Friction Setup, DEFAULT_MODEL is gpt-4o-mini, but set ANTHROPIC_API_KEY #433 (provider key documentation) — does not close
• Related to feat(config): add GLM-5.1 and GLM-5V-Turbo to GLM preset models #436 (GLM coverage) — verified GLM models are already present
• Regression coverage around server sync behavior from bug: stale modelId/providerId after server provider sync #226 — existing implementation already handles this

Checklist

• Model IDs verified against official documentation
• .env.example updated with example model lists
• No breaking changes
• Server-config extensibility already implemented (no code changes needed)

[wyuc]

Apologies @wahajahmed010 — the deepseek-v4-pro[1m] line in #480 was a mistake on my part. [1m] is only valid on DeepSeek's Anthropic-compatible endpoint (Claude Code), not the OpenAI-compatible API OpenMAIC uses; the base deepseek-v4-pro already defaults to 1M context. Closing.