[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"pack-detail-china-programmer-ai-stack-fr":3,"seo:pack:china-programmer-ai-stack:fr":103},{"code":4,"message":5,"data":6},200,"操作成功",{"pack":7},{"slug":8,"icon":9,"tone":10,"status":11,"status_label":12,"title":13,"description":14,"items":15,"install_cmd":102},"china-programmer-ai-stack","🇨🇳","#DC2626","new","Nouveau · cette semaine","Stack IA du Développeur en Chine — Coder Derrière le GFW","Dix outils pour développeurs en Chine continentale qui ne peuvent (ou ne veulent) pas dépendre de ChatGPT\u002FCursor : DeepSeek + Qwen + Kimi comme couche modèle, Cherry Studio comme client desktop, One API \u002F LiteLLM comme passerelle unifiée, Tabby \u002F Ollama pour équipes intranet isolées, plus Continue et CLIs natifs DeepSeek qui acceptent le RMB sans VPN.",[16,28,36,44,54,62,72,80,88,95],{"id":17,"uuid":18,"slug":19,"title":20,"description":21,"author_name":22,"view_count":23,"vote_count":24,"lang_type":25,"type":26,"type_label":27},3142,"d950a025-f3c2-477b-b2fb-db2975bcafd7","deepseek-tui-terminal-coding-agent-for-deepseek","DeepSeek TUI — Terminal Coding Agent for DeepSeek","DeepSeek TUI is a terminal coding agent for DeepSeek models. Install via npm\u002Fcargo\u002Fbrew, run on your repo, and keep approval gates for edits.","Script Depot",81,0,"en","script","Script",{"id":29,"uuid":30,"slug":31,"title":32,"description":33,"author_name":34,"view_count":35,"vote_count":24,"lang_type":25,"type":26,"type_label":27},3022,"2dff2d41-ba09-4f17-bcce-22b9537c1db9","qwen-code-terminal-coding-agent-for-qwen-models","Qwen Code — Terminal Coding Agent for Qwen Models","Qwen Code is an open-source terminal coding agent for Qwen models. Node 22+, npm or Homebrew install, \u002Fauth flow, codebase Q&A, refactors, and tests.","QwenLM",99,{"id":37,"uuid":38,"slug":39,"title":40,"description":41,"author_name":42,"view_count":43,"vote_count":24,"lang_type":25,"type":26,"type_label":27},3643,"b6547ac2-21f2-5e33-98db-4ddc3552d152","oh-my-kimi-evidence-gated-agent-runtime-for-kimi","oh-my-kimi — Evidence-gated Agent Runtime for Kimi","oh-my-kimi (OMK) is a CLI runtime that adds evidence gates and worktree isolation to Kimi Code; verified 69★ and ships `omk init\u002Fdoctor\u002Fchat`.","Skill Factory",104,{"id":45,"uuid":46,"slug":47,"title":48,"description":49,"author_name":50,"view_count":51,"vote_count":24,"lang_type":25,"type":52,"type_label":53},2821,"121c2d1e-678a-4ea4-8799-70391f0a11de","cherry-studio-custom-models-byok-any-llm-provider","Cherry Studio Custom Models — BYOK Any LLM Provider","Cherry Studio Custom Models adds any OpenAI-compatible endpoint — proxy, local, or third-party. Mix Claude, GPT, Gemini, DeepSeek, Ollama side-by-side.","Cherry Studio",107,"skill","Skill",{"id":55,"uuid":56,"slug":57,"title":58,"description":59,"author_name":60,"view_count":61,"vote_count":24,"lang_type":25,"type":52,"type_label":53},3821,"65f4b9a7-d932-51b8-aa22-6f0ff7cff276","one-api-unified-llm-api-gateway-docker","One API — Unified LLM API Gateway (Docker)","One API is a self-hosted LLM API gateway: unify OpenAI\u002FClaude\u002FGemini\u002FDeepSeek endpoints, manage keys, and deploy via Docker in minutes (33.7k★).","AI Open Source",89,{"id":63,"uuid":64,"slug":65,"title":66,"description":67,"author_name":68,"view_count":69,"vote_count":24,"lang_type":25,"type":70,"type_label":71},2789,"0f113965-1adc-4435-982b-fb613fa4d157","litellm-proxy-unified-gateway-for-100-llm-apis","LiteLLM Proxy — Unified Gateway for 100+ LLM APIs","LiteLLM Proxy maps 100+ LLM providers (Anthropic, OpenAI, Bedrock, Vertex) to one OpenAI-compatible endpoint. Auth, rate limit, cost track, fallbacks.","LiteLLM (BerriAI)",92,"agent","Agent",{"id":73,"uuid":74,"slug":75,"title":76,"description":77,"author_name":78,"view_count":79,"vote_count":24,"lang_type":25,"type":52,"type_label":53},613,"faed3b71-acbc-41f6-98f6-317143d41138","continue-open-source-ai-code-assistant-ides-faed3b71","Continue — Open-Source AI Code Assistant for IDEs","Open-source AI code assistant for VS Code and JetBrains. Connect any LLM model, use autocomplete, chat, and inline edits. Fully customizable with your own models and context. 22,000+ stars.","Continue",152,{"id":81,"uuid":82,"slug":83,"title":84,"description":85,"author_name":86,"view_count":87,"vote_count":24,"lang_type":25,"type":52,"type_label":53},216,"1a1d4061-a148-4566-a3d7-ab40e6f2a972","tabby-self-hosted-ai-coding-assistant-1a1d4061","Tabby — Self-Hosted AI Coding Assistant","Self-hosted AI code completion and chat assistant. Privacy-first alternative to GitHub Copilot. Supports 20+ models, repo-aware context, and IDE integrations. 33K+ stars.","TokRepo精选",908,{"id":89,"uuid":90,"slug":91,"title":92,"description":93,"author_name":22,"view_count":94,"vote_count":24,"lang_type":25,"type":52,"type_label":53},162,"0eefb7ad-754e-4f35-8967-586ebf4c2a6a","ollama-run-llms-locally-0eefb7ad","Ollama — Run LLMs Locally","Run large language models locally on your machine. Supports Llama 3, Mistral, Gemma, Phi, and dozens more. One-command install, OpenAI-compatible API.",197,{"id":96,"uuid":97,"slug":98,"title":99,"description":100,"author_name":22,"view_count":101,"vote_count":24,"lang_type":25,"type":26,"type_label":27},3604,"42bb7d1b-7879-54af-b319-83ecb7d9b53a","reasonix-deepseek-native-coding-agent-cli","Reasonix — DeepSeek-Native Coding Agent CLI","Reasonix is a DeepSeek-native coding agent CLI for the terminal; README reports a 99.82% cache hit in a real-day case study.",112,"tokrepo install pack\u002Fchina-programmer-ai-stack",{"pageType":104,"pageKey":8,"locale":25,"title":105,"metaDescription":106,"h1":107,"tldr":108,"bodyMarkdown":109,"faq":110,"schema":126,"internalLinks":132,"citations":145,"wordCount":158,"generatedAt":159},"pack","Chinese Programmer's China Stack — 10 AI Coding Tools That Work Behind the GFW","DeepSeek + Qwen + Kimi for models, Cherry Studio as the desktop chat shell, One API + LiteLLM for unified gateways, Tabby + Ollama for air-gapped intranets, plus Continue and DeepSeek-native CLIs. The China-realistic AI coding stack when Cursor and Codex don't work. Install order and gotchas included.","Chinese Programmer's China Stack — Real AI Coding Without ChatGPT or Cursor","Ten picks for engineers in mainland China who can't reliably reach ChatGPT, Claude, or Cursor: pick a domestic model (DeepSeek \u002F Qwen \u002F Kimi) that takes RMB and needs no VPN, wrap it in Cherry Studio for chat and One API for gateway, drop in Tabby or Ollama if your network is fully air-gapped, and use Continue or a DeepSeek-native CLI for the IDE. Install order matters.","## What's in this pack\n\nThis is the rig for a working engineer in **mainland China** who hits the same three walls every week: **Cursor and Codex sign-up requires a foreign phone**, **OpenAI \u002F Anthropic APIs are intermittently blocked at the network layer**, and **corporate firewalls + WeChat-Pay-only billing make foreign SaaS a non-starter**. The pack is not a workaround for those tools — it's the **native-China alternative** built on Chinese model labs (DeepSeek, Alibaba Qwen, Moonshot Kimi) plus self-hostable gateways and IDE plugins.\n\nEvery pick here meets three criteria: **RMB-payable or self-hostable**, **reachable from a typical mainland network without a VPN**, and **good enough to actually use as a daily coding assistant** — not just \"works\". If you have Cursor working and don't want to switch, this isn't your pack. If you're on a company-issued laptop with no foreign network access, this is the only pack that ships.\n\n## Install in this order\n\n1. **DeepSeek TUI** *(id: 3142)* — start here because **DeepSeek-V3 \u002F R1 is the highest-quality Chinese-trained model with an RMB-billable API**, and the TUI is a one-binary terminal agent that proves the API key works before you wire anything else up. `npm i -g deepseek-tui`, paste your key from `platform.deepseek.com`, get answers in 30 seconds.\n2. **Qwen Code** *(id: 3022)* — Alibaba's official terminal coding agent for Qwen models. **Install second** because Qwen-Max \u002F Qwen3-Coder are the strongest alternative when DeepSeek capacity is throttled, and you want a fallback model wired up before you write any production scripts. `npm i -g @qwen-code\u002Fqwen-code`, `\u002Fauth` with your DashScope (bailian.console.aliyun.com) key.\n3. **oh-my-kimi** *(id: 3643)* — Moonshot's Kimi K2 has the **longest reliable context window** (1M tokens) among China-hosted models, and oh-my-kimi gates the agent loop with evidence checks. **Install third** to round out the model layer: DeepSeek for daily code, Qwen as backup, Kimi for whole-codebase questions.\n4. **Cherry Studio Custom Models** *(id: 2821)* — open-source desktop chat client (Mac\u002FWin\u002FLinux) that **adds any OpenAI-compatible endpoint as a BYOK provider**. Install fourth because once you have three model providers, you need one chat UI that can flip between them mid-conversation. Cherry Studio is the de facto standard in Chinese dev communities — 30K+ stars, MIT license, no telemetry.\n5. **One API** *(id: 3821)* — self-hosted LLM API gateway (Docker, Go). **Install fifth** because by now you have multiple keys (DeepSeek + Qwen + Kimi) and your team needs a single OpenAI-compatible URL that round-robins across them. One API is the gateway most Chinese teams reach for first (33K+ stars, Chinese-language admin UI).\n6. **LiteLLM Proxy** *(id: 2789)* — the international alternative gateway, Python-based, 100+ provider support. Install **as an alternative to** One API only if you also need to proxy international providers (OpenRouter, Together, Anthropic via reverse-proxy) and want richer cost tracking. Don't run both.\n7. **Continue** *(id: 613)* — open-source IDE assistant (VSCode + JetBrains) that **points at any OpenAI-compatible endpoint** — which means it points at your One API gateway, which means your IDE autocomplete and chat are now powered by DeepSeek-Coder for ¥0.50 per million tokens.\n8. **Tabby** *(id: 216)* — self-hosted GitHub Copilot alternative. **Install only if your company forbids external API calls** (banks, gov, internal R&D). Tabby runs entirely on your hardware, supports DeepSeek-Coder \u002F Qwen-Coder weights, and gives FIM autocomplete in VSCode + JetBrains.\n9. **Ollama** *(id: 162)* — local model runtime. **Install if you have a Mac with 32GB+ unified memory or a workstation with a 24GB+ GPU**. Run `ollama pull qwen3:14b` or `deepseek-r1:7b` and Cherry Studio \u002F Continue can talk to `localhost:11434` instead of any API.\n10. **Reasonix** *(id: 3604)* — DeepSeek-native coding agent CLI optimized for prompt caching (claimed 99%+ cache hit). **Install last** as the power-user upgrade once you've used DeepSeek TUI for a week and want lower latency + lower bills on a real repo.\n\n## How they fit together\n\n```\n  Domestic model APIs (all RMB-billable, no VPN needed)\n  ┌─────────────────┬─────────────────┬─────────────────┐\n  │ DeepSeek-V3\u002FR1  │  Qwen-Max\u002FCoder │  Kimi K2 (1M ctx)│\n  │ (deepseek.com)  │  (bailian.aliyun)│ (platform.moonshot)│\n  └────────┬────────┴────────┬────────┴────────┬─────────┘\n           │                 │                 │\n           ▼                 ▼                 ▼\n           One API  \u002F  LiteLLM Proxy  (Docker, your VPS)\n           (unified OpenAI-compat URL + key rotation + cost log)\n                              │\n        ┌─────────────────────┼─────────────────────┐\n        ▼                     ▼                     ▼\n  DeepSeek TUI \u002F         Cherry Studio        Continue (VSCode\u002FJB)\n  Qwen Code \u002F           (desktop chat,         (in-IDE chat\n  oh-my-kimi \u002F          BYOK any provider)     + completions)\n  Reasonix\n  (terminal agents)\n\n  ── Air-gapped fallback (no external network at all) ──\n          Tabby (self-hosted Copilot, runs DeepSeek-Coder weights)\n          Ollama (local Qwen \u002F DeepSeek-Coder \u002F Yi \u002F GLM weights)\n```\n\nThe inflection point is **One API in front of three domestic providers**. Before that, you have three API keys scattered across `.env` files and every tool has to know about each provider. After that, every tool points at one URL, you swap models by renaming the model field, and your billing team sees one consolidated cost report. Don't skip the gateway — without it, multi-model workflows fall apart by week two.\n\n## Tradeoffs you'll hit\n\n- **DeepSeek vs Qwen vs Kimi** — DeepSeek wins on code quality + price (¥0.50-2\u002FM tokens). Qwen wins on Chinese-language documentation, multi-modal, and Alibaba Cloud enterprise compliance. Kimi wins on context length (1M, actually usable). Most teams run DeepSeek default, Qwen as Aliyun-compliance fallback, Kimi for codebase-wide Q&A.\n- **One API vs LiteLLM** — One API is Chinese-developer-first (UI in Chinese, supports WeChat Pay billing plugins, lighter footprint). LiteLLM is international-first (better cost tracking, more provider integrations, Python ecosystem). If your only providers are Chinese, One API is the obvious pick.\n- **Continue vs DeepSeek TUI \u002F Qwen Code** — Continue lives in your IDE (autocomplete + chat panel). The CLI agents live in terminal and can execute multi-step refactors. They complement each other: Continue for typing, CLI agents for \"refactor this whole module\". Most engineers run both.\n- **Cloud API vs Tabby vs Ollama** — Cloud API is cheapest and highest quality but requires external network. Tabby is mid-quality (model is local) and needs a GPU. Ollama is your laptop; quality drops sharply below the 14B parameter mark. Only go local if compliance demands it or you genuinely have a 4090 \u002F M-series Pro burning idle cycles.\n- **Cherry Studio vs the model labs' web chat** — DeepSeek, Qwen, and Kimi all have free web chat. Cherry Studio is worth installing the moment you want to **switch models mid-conversation**, save chat history locally, or pipe in your own RAG knowledge base. If you only use one model from a browser, you don't need it.\n\n## Common pitfalls\n\n- **\"DeepSeek API is down\"** — DeepSeek has had multi-hour capacity issues during peak periods. This is exactly why One API + a Qwen fallback exists. Configure the gateway with `priority: 1 deepseek, priority: 2 qwen` and most outages become invisible.\n- **JetBrains Continue plugin marketplace can be slow from CN** — if the plugin fails to install from inside an IDE, download the `.zip` from `plugins.jetbrains.com` (often reachable) and install from disk. VSCode marketplace is generally fine via Microsoft's CN mirrors.\n- **Bailian (Aliyun) auth flow assumes a Chinese phone** — Qwen API via bailian.console.aliyun.com requires Aliyun account + Alipay real-name verification. If your team is on a foreign Aliyun account, use **Qwen via DashScope international** (`dashscope-intl.aliyuncs.com`) — different endpoint, same model, different billing entity.\n- **Cherry Studio + corporate proxy** — Cherry Studio respects system proxy by default, which breaks DeepSeek calls if your proxy whitelist only includes foreign domains. Either add `api.deepseek.com` to the proxy bypass list, or configure Cherry Studio's per-provider proxy override.\n- **Tabby + GPU memory** — DeepSeek-Coder-V2-Lite-16B needs ~24GB VRAM at fp16, ~12GB at 4-bit quant. A single 3090 \u002F 4090 is enough; a 4060Ti 16GB is not. Right-size before promising your team Copilot replacement.\n- **One API on a public IP without auth = key theft** — One API admin UI defaults to `root\u002F123456`. Change it the first hour, put it behind nginx + basic auth or a VPN. Several Chinese dev forums have screenshots of stolen DeepSeek keys racking up ¥10K bills overnight.\n- **WeChat Pay \u002F Alipay billing topup minimums** — DeepSeek and Moonshot have ¥10-50 minimum top-ups; not a problem for individuals, but for **company finance teams that need fapiao (发票)**, request the developer\u002Fenterprise account from day one — switching mid-stream wastes a week.",[111,114,117,120,123],{"q":112,"a":113},"I can use Cursor \u002F Claude with a VPN — why bother switching to a domestic stack?","Three reasons: (1) your company laptop and corporate VPN will eventually conflict with personal VPN tools, and the day you can't log in to a customer call is the day this stack pays for itself; (2) DeepSeek and Qwen are now genuinely competitive on code quality for ~1\u002F10 the price — even with stable foreign access, the cost difference matters at team scale; (3) fapiao (发票). If your company finance team can't reimburse a USD Stripe charge, RMB-billable domestic models are the only practical path. Most engineers end up running both — foreign tools when convenient, this stack as the always-works fallback.",{"q":115,"a":116},"My company forbids any external API call — what subset of this pack works?","Drop the cloud APIs entirely and run: Tabby (self-hosted Copilot, item 8) + Ollama (local model runtime, item 9) + Continue (IDE plugin, item 7) pointed at Ollama's localhost endpoint. Cherry Studio (item 4) also works pointed at a local model. You'll need a workstation with a 24GB+ GPU or an M-series Mac with 32GB+ unified memory. Model recommendations: DeepSeek-Coder-V2-Lite-16B for code, Qwen3-14B for general chat. Quality is below cloud DeepSeek but well above no-AI.",{"q":118,"a":119},"How much does this stack actually cost per engineer per month?","Individual use with DeepSeek as primary: ¥30-80\u002Fmonth for heavy daily use (1-3M tokens\u002Fday). With One API doing fallback to Qwen, add another ¥20-40. Compare to Cursor Pro at ~$20\u002Fmonth (~¥150) plus Copilot at $10\u002Fmonth (~¥75) — and that's assuming the VPN works. For a team of 10, a self-hosted One API on a ¥80\u002Fmonth VPS plus pooled DeepSeek+Qwen keys typically lands at ¥500-1500\u002Fmonth total versus ¥6000+ for international SaaS equivalents.",{"q":121,"a":122},"Will my company finance team accept the bills (need fapiao 发票)?","DeepSeek, Alibaba (Qwen via bailian), and Moonshot (Kimi) all issue Chinese VAT fapiao (普票 \u002F 专票) on request — request from day one and register the enterprise account, not the individual. One API and LiteLLM are open-source self-hosted, so there's no SaaS bill to invoice. The catch: switching from a personal Aliyun account to an enterprise account mid-month is painful (separate billing entities). Set up the enterprise account first, then create developer keys under it.",{"q":124,"a":125},"Which model should I use for what task?","Default for code: DeepSeek-V3 (chat) + DeepSeek-Coder (autocomplete) — best quality-per-yuan in 2026. Reasoning \u002F hard math \u002F refactor planning: DeepSeek-R1 or DeepSeek-V3 thinking mode. Whole-codebase Q&A or document analysis (>200K tokens): Kimi K2 — its 1M context actually works (most others degrade past 64K). Multi-modal \u002F vision \u002F Aliyun-cloud-native enterprise: Qwen-VL or Qwen-Max. Air-gapped local: DeepSeek-Coder-V2-Lite-16B (code) or Qwen3-14B (general). Don't pick one model and stick with it — flipping between providers via Cherry Studio or One API is the whole point.",{"@context":127,"@type":128,"name":129,"description":130,"numberOfItems":131,"inLanguage":25},"https:\u002F\u002Fschema.org","ItemList","Chinese Programmer's China Stack","Ten AI coding tools curated for developers in mainland China who can't (or won't) rely on ChatGPT, Claude, or Cursor — domestic models, self-hostable gateways, and IDE plugins that take RMB and need no VPN.",10,[133,137,141],{"url":134,"anchor":135,"reason":136},"\u002Fen\u002Fpacks\u002Fwindows-engineer-ai-stack","Windows Engineer's AI Stack","Many CN developers run Windows + WSL — this pack pairs with the China stack for the OS layer",{"url":138,"anchor":139,"reason":140},"\u002Fen\u002Fpacks\u002Fterminal-first-workflow-ai","Terminal-First AI Workflow","The CLI agents in this pack (DeepSeek TUI, Qwen Code, Reasonix) live in a broader terminal-first ecosystem",{"url":142,"anchor":143,"reason":144},"\u002Fen\u002Fpacks\u002Faider-pair-programming","Aider Pair Programming","Aider is a model-agnostic alternative that pairs well with DeepSeek-Coder via One API",[146,150,154],{"claim":147,"source_name":148,"source_url":149},"DeepSeek-V3 is an open-weight MoE model with public RMB-billable API","DeepSeek platform docs","https:\u002F\u002Fplatform.deepseek.com\u002F",{"claim":151,"source_name":152,"source_url":153},"Qwen Code is the official terminal coding agent from QwenLM with Apache-2.0 license","QwenLM\u002Fqwen-code on GitHub","https:\u002F\u002Fgithub.com\u002FQwenLM\u002Fqwen-code",{"claim":155,"source_name":156,"source_url":157},"One API is an open-source self-hosted LLM gateway with 33K+ GitHub stars","One API on GitHub","https:\u002F\u002Fgithub.com\u002Fsongquanpeng\u002Fone-api",1180,"2026-05-23T12:00:00Z"]