{"skill":{"slug":"chinese-llm-router","displayName":"Chinese LLM Router","summary":"Route OpenClaw chats to top Chinese LLMs with smart model selection, auto-fallback, cost tracking, and unified OpenAI-compatible API access.","description":"# Chinese LLM Router\n\nRoute your OpenClaw conversations to the best Chinese AI models — no config headaches, just pick and chat.\n\n## What It Does\n\nGives your OpenClaw instant access to all major Chinese LLMs through a single unified interface:\n\n- **DeepSeek** (V3.2 / R1) — Best open-source reasoning, dirt cheap\n- **Qwen** (Qwen3-Max / Qwen3-Max-Thinking / Qwen3-Coder-Plus) — Alibaba's flagship, strong all-rounder\n- **GLM** (GLM-5 / GLM-4.7) — Zhipu AI, top-tier coding & agent tasks\n- **Kimi** (K2.5 / K2.5-Thinking) — Moonshot AI, great for long context & vision\n- **Doubao Seed 2.0** (Pro / Lite / Mini) — ByteDance, fast & cheap\n- **MiniMax** (M2.5) — Lightweight powerhouse, runs locally too\n- **Step** (3.5 Flash) — StepFun, blazing fast inference\n- **Baichuan** (Baichuan4-Turbo) — Strong Chinese language understanding\n- **Spark** (v4.0 Ultra) — iFlytek, speech & Chinese NLP specialist\n- **Hunyuan** (Turbo-S) — Tencent, WeChat ecosystem integration\n\n## Quick Start\n\nTell your OpenClaw:\n\n```\nUse DeepSeek V3.2 for this conversation\n```\n\nOr ask it to pick the best model:\n\n```\nWhich Chinese model is best for coding? Switch to it.\n```\n\n## Commands\n\n| Command | What it does |\n|---------|-------------|\n| `list models` | Show all available Chinese LLMs with status |\n| `use <model>` | Switch to a specific model |\n| `compare <models>` | Compare capabilities & pricing |\n| `recommend <task>` | Get model recommendation for a task type |\n| `test <model>` | Send a test prompt to verify connectivity |\n| `status` | Check which models are currently accessible |\n\n## Model Selection Guide\n\n| Task | Recommended Model | Why |\n|------|------------------|-----|\n| General chat | Qwen3-Max | Best all-rounder, strong Chinese |\n| Coding | GLM-5 / Kimi K2.5 | Top coding benchmarks |\n| Math & reasoning | DeepSeek R1 | Purpose-built for reasoning |\n| Long documents | Kimi K2.5 (128K) / DeepSeek V3.2 (1M) | Massive context windows |\n| Fast & cheap | Step 3.5 Flash / Doubao Seed 2.0 Mini | Sub-second latency |\n| Creative writing | Qwen3-Max / Doubao Seed 2.0 Pro | Rich Chinese expression |\n| Agent tasks | GLM-5 / Qwen3-Max | Best tool-use support |\n\n## Configuration\n\nThe skill reads API keys from environment or from `~/.chinese-llm-router/config.json`:\n\n```json\n{\n  \"providers\": {\n    \"deepseek\": {\n      \"apiKey\": \"sk-xxx\",\n      \"baseUrl\": \"https://api.deepseek.com/v1\",\n      \"models\": [\"deepseek-chat\", \"deepseek-reasoner\"]\n    },\n    \"qwen\": {\n      \"apiKey\": \"sk-xxx\",\n      \"baseUrl\": \"https://dashscope.aliyuncs.com/compatible-mode/v1\",\n      \"models\": [\"qwen3-max\", \"qwen3-max-thinking\", \"qwen3-coder-plus\"]\n    },\n    \"glm\": {\n      \"apiKey\": \"xxx.xxx\",\n      \"baseUrl\": \"https://open.bigmodel.cn/api/paas/v4\",\n      \"models\": [\"glm-5\", \"glm-4-plus\"]\n    },\n    \"kimi\": {\n      \"apiKey\": \"sk-xxx\",\n      \"baseUrl\": \"https://api.moonshot.cn/v1\",\n      \"models\": [\"kimi-k2.5\", \"kimi-k2.5-thinking\"]\n    },\n    \"doubao\": {\n      \"apiKey\": \"xxx\",\n      \"baseUrl\": \"https://ark.cn-beijing.volces.com/api/v3\",\n      \"models\": [\"doubao-seed-2.0-pro\", \"doubao-seed-2.0-lite\", \"doubao-seed-2.0-mini\"]\n    },\n    \"minimax\": {\n      \"apiKey\": \"xxx\",\n      \"baseUrl\": \"https://api.minimax.chat/v1\",\n      \"models\": [\"minimax-m2.5\"]\n    },\n    \"step\": {\n      \"apiKey\": \"xxx\",\n      \"baseUrl\": \"https://api.stepfun.com/v1\",\n      \"models\": [\"step-3.5-flash\"]\n    },\n    \"baichuan\": {\n      \"apiKey\": \"xxx\",\n      \"baseUrl\": \"https://api.baichuan-ai.com/v1\",\n      \"models\": [\"baichuan4-turbo\"]\n    },\n    \"spark\": {\n      \"apiKey\": \"xxx\",\n      \"baseUrl\": \"https://spark-api-open.xf-yun.com/v1\",\n      \"models\": [\"spark-v4.0-ultra\"]\n    },\n    \"hunyuan\": {\n      \"apiKey\": \"xxx\",\n      \"baseUrl\": \"https://api.hunyuan.cloud.tencent.com/v1\",\n      \"models\": [\"hunyuan-turbo-s\"]\n    }\n  },\n  \"default\": \"qwen3-max\",\n  \"fallback\": [\"deepseek-chat\", \"doubao-seed-2.0-pro\"]\n}\n```\n\n## Setup\n\n1. Get API keys from the providers you want (most offer free tiers):\n   - DeepSeek: https://platform.deepseek.com\n   - Qwen (Alibaba): https://dashscope.console.aliyun.com\n   - GLM (Zhipu): https://open.bigmodel.cn\n   - Kimi (Moonshot): https://platform.moonshot.cn\n   - Doubao (ByteDance): https://console.volcengine.com/ark\n   - MiniMax: https://platform.minimaxi.com\n   - Step (StepFun): https://platform.stepfun.com\n   - Baichuan: https://platform.baichuan-ai.com\n   - Spark (iFlytek): https://console.xfyun.cn\n   - Hunyuan (Tencent): https://cloud.tencent.com/product/hunyuan\n\n2. Run the setup script:\n   ```bash\n   node scripts/setup.js\n   ```\n\n3. Done! Your OpenClaw can now use any configured model.\n\n## Pricing Reference (Feb 2026)\n\n| Model | Input (¥/M tokens) | Output (¥/M tokens) | Notes |\n|-------|-------------------|---------------------|-------|\n| DeepSeek V3.2 | ¥0.5 (cache ¥0.1) | ¥2.0 | Cheapest flagship |\n| Qwen3-Max | ¥2.0 | ¥6.0 | Free tier available |\n| GLM-5 | ¥5.0 | ¥5.0 | Just launched, may change |\n| Kimi K2.5 | ¥2.0 | ¥6.0 | Open source, self-host free |\n| Doubao Seed 2.0 Pro | ¥0.8 | ¥2.0 | ByteDance subsidy |\n| Doubao Seed 2.0 Mini | ¥0.15 | ¥0.3 | Ultra cheap |\n| MiniMax M2.5 | ¥1.0 | ¥3.0 | Can run locally |\n| Step 3.5 Flash | ¥0.7 | ¥1.4 | Fastest inference |\n\n*Prices as of Feb 2026. All providers offer free tiers or credits for new users.*\n\n## All APIs are OpenAI-Compatible\n\nEvery provider listed uses the OpenAI chat/completions format. No special SDKs needed — just change `baseUrl` and `apiKey`.\n\n## Features\n\n- **Auto-fallback**: If one provider is down, automatically try the next\n- **Cost tracking**: See per-model token usage and estimated cost\n- **Smart routing**: Describe your task, get the best model recommendation\n- **Batch compare**: Send the same prompt to multiple models, compare outputs\n- **Context-aware**: Remembers your model preference per conversation topic\n\n## Links\n\n- 🦐 Try our AI Plaza: https://ai.xudd-v.com\n- 📦 ClawHub: https://clawhub.ai/Xdd-xund/chinese-llm-router\n- 💬 Feedback: https://ai.xudd-v.com/connect.html\n","tags":{"chinese":"1.0.0","deepseek":"1.0.0","doubao":"1.0.0","glm":"1.0.0","kimi":"1.0.0","latest":"1.0.0","llm":"1.0.0","minimax":"1.0.0","qwen":"1.0.0","router":"1.0.0","step":"1.0.0"},"stats":{"comments":0,"downloads":1561,"installsAllTime":59,"installsCurrent":3,"stars":0,"versions":1},"createdAt":1771262128322,"updatedAt":1778991001536},"latestVersion":{"version":"1.0.0","createdAt":1771262128322,"changelog":"Initial release of Chinese LLM Router.\n\n- Unified interface to access all major Chinese LLMs in OpenClaw.\n- Supports model selection, comparison, recommendations, and testing via commands.\n- Auto-fallback, cost tracking, and smart routing between multiple providers.\n- Simple configuration with OpenAI-compatible APIs and quick setup script.\n- Comprehensive model/task recommendations and up-to-date pricing included.","license":null},"metadata":null,"owner":{"handle":"xdd-xund","userId":"s1772a884c28sndtzdrcv0sr31885nwb","displayName":"xund","image":"https://avatars.githubusercontent.com/u/194621946?v=4"},"moderation":null}