{"slug": "qwen-code", "name": "Qwen Code", "category": "ai-coding", "category_name": "AI coding tools", "description": "", "recommenders": 6, "commenters": 10, "warners": 0, "score": 5.073, "rising": -0.64, "first": "2025-11-11", "latest": "2026-10-08", "why": "Together with Qwen Code (fork of Gemini) it is actually pretty capable.", "why_is_quote": true, "has_verdict": false, "sentiment": {"recommend": 6, "neutral": 5, "compare": 0, "warn": 0}, "spark": [1, 0, 1, 1, 0, 1, 1, 0, 0, 0, 1, 0], "listed": true, "loved": [], "caveat": "", "short_quote": "Together with Qwen Code (fork of Gemini) it is actually pretty capable.", "kind": "", "icon": true, "domain": null, "github_repo": null, "official_url": "https://github.com/QwenLM/qwen-code", "pricing": "", "aliases": ["Qwen Code", "Qwen CLI"], "verdict": null, "praise": [{"label": "", "summary": "", "quotes": [{"id": 795782640, "item_id": 49736415, "url": "https://news.ycombinator.com/item?id=49736415", "by": "jonsoft", "date": "2026-09-17", "version": null, "stance": "recommend", "quote": "I made this 3D game in a day on the same setup with Qwen Code as agent", "excerpt": "I made this 3D game in a day on the same setup with Qwen Code as agent: https://games.jonathanpage.com/ And I am not a web developer! It's an extraordinary model. (Mouse and keyboard required)", "context": "", "use_case": "building a 3D game", "switched_from": "", "story": "Xiaomi Mimo 2.6 live post-training dashboard"}, {"id": 768792739, "item_id": 48049546, "url": "https://news.ycombinator.com/item?id=48049546", "by": "neutrinobro", "date": "2026-05-07", "version": null, "stance": "recommend", "quote": "I have been using the qwen-code harness though", "excerpt": "I have a old supermicro X10DRU-i server with two Tesla V100's (48 GB VRAM) and 128GB RAM and have been running qwen3.6-27B with a lot of success. I would say it's performance on my use case (modifying and extending a ~70kloc C++ code base) has been excellent. I have no benchmarks, but it seems comparable to claude sonnet 4.6 in capabilities. I run it with llama.cpp: llama-server -m Qwen3.6-27B-Q8_0.gguf -c 131072 --t", "context": "", "use_case": "", "switched_from": "", "story": "ZAYA1-8B: An 8B Moe Model with 760M Active Params Matching DeepSeek-R1 on Math"}, {"id": 764707376, "item_id": 47794211, "url": "https://news.ycombinator.com/item?id=47794211", "by": "alberto-m", "date": "2026-04-16", "version": null, "stance": "recommend", "quote": "it was powerful enough to complete 95% of a small hobby project", "excerpt": "I used Qwen CLI's undescribed \u201ccoder_agent\u201d (I guess Qwen 3.5 with size auto-selection) and it was powerful enough to complete 95% of a small hobby project involving coding, reverse engineering and debugging. Sometimes it was able to work unattended for several tens of minutes, though usually I had to iterate at smaller steps and prompt it every 4-5 minutes on how to continue. I'd rate it a little below the top model", "context": "", "use_case": "", "switched_from": "", "story": "Qwen3.6-35B-A3B: Agentic Coding Power, Now Open to All"}, {"id": 750358865, "item_id": 46897429, "url": "https://news.ycombinator.com/item?id=46897429", "by": "dust42", "date": "2026-02-05", "version": null, "stance": "recommend", "quote": "Together with Qwen Code (fork of Gemini) it is actually pretty capable.", "excerpt": "The brand new Qwen3-Coder-Next runs at 300Tok/s PP and 40Tok/s on M1 64GB with 4-bit MLX quant. Together with Qwen Code (fork of Gemini) it is actually pretty capable. Before that I used Qwen3-30B which is good enough for some quick javascript or Python, like 'add a new endpoint /api/foobar which does foobaz'. Also very decent for a quick summary of code. It is 530Tok/s PP and 50Tok/s TG. If you have it spit out lots", "context": "", "use_case": "local coding agent", "switched_from": "", "story": "Claude Code: connect to a local model when your quota runs out"}, {"id": 747061954, "item_id": 46691372, "url": "https://news.ycombinator.com/item?id=46691372", "by": "mx-Liu123", "date": "2026-01-20", "version": null, "stance": "recommend", "quote": "Qwen CLI is a great alternative if you want an OpenAI-compatible API with a generous free tier", "excerpt": "Author's Note: A few technical details for those looking to try AgentCommander: Why Gemini/Qwen CLI?: I chose these as backends because they offer robust directory isolation. I tried integrating Claude Code, but found it difficult to restrict its file-system reach. Qwen CLI is a great alternative if you want an OpenAI-compatible API with a generous free tier (2,000 requests/day). Environment: Ensure you have Python 3", "context": "", "use_case": "", "switched_from": "", "story": "Show HN: AgentCommander - workflow engine for evolutionary code optimization"}]}], "caveats": [], "who": [], "tags": [], "use_cases": ["coding agent", "building a 3d game", "running qwen", "local coding agent", "coding agent with swappable models"], "compared": [{"slug": "qwen", "name": "Qwen", "co": 10, "switched_from": 0, "switched_to": 0}, {"slug": "claude-code", "name": "Claude Code", "co": 6, "switched_from": 0, "switched_to": 0}, {"slug": "gemini-cli", "name": "Gemini CLI", "co": 3, "switched_from": 0, "switched_to": 0}, {"slug": "opencode", "name": "OpenCode", "co": 2, "switched_from": 0, "switched_to": 0}, {"slug": "claude-fable", "name": "Claude Fable", "co": 2, "switched_from": 0, "switched_to": 0}, {"slug": "claude-sonnet", "name": "Claude Sonnet", "co": 2, "switched_from": 0, "switched_to": 0}, {"slug": "pi", "name": "Pi", "co": 2, "switched_from": 0, "switched_to": 0}], "makers": [], "threads": [{"url": "https://news.ycombinator.com/item?id=46691343", "title": "Show HN: AgentCommander - workflow engine for evolutionary code optimization", "mentions": 2}, {"url": "https://news.ycombinator.com/item?id=50000488", "title": "Why isn't the industry freaking out about DeepSeek 4.1 Flash?", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=49732270", "title": "Xiaomi Mimo 2.6 live post-training dashboard", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=49150470", "title": "Qwen3.8-Max: A New Bar for Coding and Cowork", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=48999291", "title": "Kimi K3 Is Competitive with Fable; Kimi K3 and Fable Is SoTA", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=48883275", "title": "Claude Code sends 33k tokens before reading the prompt; OpenCode sends 7k", "mentions": 1}], "sentiment_by_month": {"2025-11": {"recommend": 1}, "2025-12": {}, "2026-01": {"recommend": 1}, "2026-02": {"recommend": 1}, "2026-03": {}, "2026-04": {"recommend": 1, "neutral": 1}, "2026-05": {"recommend": 1}, "2026-06": {}, "2026-07": {"neutral": 2}, "2026-08": {"neutral": 1}, "2026-09": {"recommend": 1}, "2026-10": {"neutral": 1}}, "versions": [], "version_unnamed": 0, "family_ref": null, "runs_on": null, "lines": [], "quotes": [{"id": 800208339, "item_id": 50013021, "url": "https://news.ycombinator.com/item?id=50013021", "by": "jonsoft", "date": "2026-10-08", "version": null, "stance": "neutral", "quote": "using Qwen Code as agent", "excerpt": "DeepSeek V4 Flash runs well on two Sparks, I documented that here: https://blog.jonathanpage.com/ GLM 5.3 Flash runs fine on two Sparks and Qwen 3.8 Flash Next on one is indeed incredible! I made this 3D game with it in two days using Qwen Code as agent: https://games.jonathanpage.com/", "context": "", "use_case": "", "switched_from": "", "story": "Why isn't the industry freaking out about DeepSeek 4.1 Flash?"}, {"id": 795782640, "item_id": 49736415, "url": "https://news.ycombinator.com/item?id=49736415", "by": "jonsoft", "date": "2026-09-17", "version": null, "stance": "recommend", "quote": "I made this 3D game in a day on the same setup with Qwen Code as agent", "excerpt": "I made this 3D game in a day on the same setup with Qwen Code as agent: https://games.jonathanpage.com/ And I am not a web developer! It's an extraordinary model. (Mouse and keyboard required)", "context": "", "use_case": "building a 3D game", "switched_from": "", "story": "Xiaomi Mimo 2.6 live post-training dashboard"}, {"id": 786495121, "item_id": 49155945, "url": "https://news.ycombinator.com/item?id=49155945", "by": "nerdalytics", "date": "2026-08-03", "version": null, "stance": "neutral", "quote": "I used Qwen together with Qwen Code.", "excerpt": "I used Qwen3.8 Max Preview for 2 weeks and can't tell if I truly miss Fable. Qwen doesn't overuse em-dashes, weird sentence structures with bold text: normal text. In my private GitHub Repo Qwen finished as much tasks as Fable did. Without hitting 5-hour session limits. Qwen's token-per-second performance fluctuated greatly. From 20 tps up to 80 tps. Qwen failed on some \"reading between the lines\", but so did Fable. ", "context": "", "use_case": "coding agent", "switched_from": "", "story": "Qwen3.8-Max: A New Bar for Coding and Cowork"}, {"id": 784074804, "item_id": 49004675, "url": "https://news.ycombinator.com/item?id=49004675", "by": "senko", "date": "2026-07-22", "version": null, "stance": "neutral", "quote": "For Qwen I used Qwen Code", "excerpt": "For the web app task I mentioned: * Kimi K3: 9532k input (9172k cached), 114k output - cost $5.5 * Qwen 3.8 Max: 18020k input (17836k cached), 114k output - cost $6.3 * Fable: ~14m input (all cached??), 196k output - cost $30 Correction on my earlier post, Kimi was through Pi, not Kimi Code. For Qwen I used Qwen Code and for Fable I used Claude Code. Not sure wtf is going on with the Fable stats (a lot tokens, virtua", "context": "", "use_case": "running Qwen", "switched_from": "", "story": "Kimi K3 Is Competitive with Fable; Kimi K3 and Fable Is SoTA"}, {"id": 782152257, "item_id": 48884516, "url": "https://news.ycombinator.com/item?id=48884516", "by": "docheinestages", "date": "2026-07-12", "version": null, "stance": "neutral", "quote": "it often fails during `write_file` tool calls, even with Qwen Code.", "excerpt": "I'm on a 48 GB M5 Macbook Pro. I use 4-bit quants with a context window of 16-32k. I tried Qwen 3.6 27B, but I can only get around 10 tokens per second, but it's painfully slow, and it often fails during `write_file` tool calls, even with Qwen Code.", "context": "", "use_case": "coding agent", "switched_from": "", "story": "Claude Code sends 33k tokens before reading the prompt; OpenCode sends 7k"}, {"id": 768792739, "item_id": 48049546, "url": "https://news.ycombinator.com/item?id=48049546", "by": "neutrinobro", "date": "2026-05-07", "version": null, "stance": "recommend", "quote": "I have been using the qwen-code harness though", "excerpt": "I have a old supermicro X10DRU-i server with two Tesla V100's (48 GB VRAM) and 128GB RAM and have been running qwen3.6-27B with a lot of success. I would say it's performance on my use case (modifying and extending a ~70kloc C++ code base) has been excellent. I have no benchmarks, but it seems comparable to claude sonnet 4.6 in capabilities. I run it with llama.cpp: llama-server -m Qwen3.6-27B-Q8_0.gguf -c 131072 --t", "context": "", "use_case": "", "switched_from": "", "story": "ZAYA1-8B: An 8B Moe Model with 760M Active Params Matching DeepSeek-R1 on Math"}, {"id": 765372081, "item_id": 47835755, "url": "https://news.ycombinator.com/item?id=47835755", "by": "atilimcetin", "date": "2026-04-20", "version": null, "stance": "neutral", "quote": "planning to switch to Qwen Code with subscription.", "excerpt": "I'm directly using https://chat.qwen.ai (Qwen3.6-Plus) and planning to switch to Qwen Code with subscription.", "context": "", "use_case": "", "switched_from": "", "story": "Qwen3.6-Max-Preview: Smarter, Sharper, Still Evolving"}, {"id": 764707376, "item_id": 47794211, "url": "https://news.ycombinator.com/item?id=47794211", "by": "alberto-m", "date": "2026-04-16", "version": null, "stance": "recommend", "quote": "it was powerful enough to complete 95% of a small hobby project", "excerpt": "I used Qwen CLI's undescribed \u201ccoder_agent\u201d (I guess Qwen 3.5 with size auto-selection) and it was powerful enough to complete 95% of a small hobby project involving coding, reverse engineering and debugging. Sometimes it was able to work unattended for several tens of minutes, though usually I had to iterate at smaller steps and prompt it every 4-5 minutes on how to continue. I'd rate it a little below the top model", "context": "", "use_case": "", "switched_from": "", "story": "Qwen3.6-35B-A3B: Agentic Coding Power, Now Open to All"}, {"id": 753503681, "item_id": 47093980, "url": "https://news.ycombinator.com/item?id=47093980", "by": "dust42", "date": "2026-02-20", "version": null, "stance": "neutral", "quote": "Qwen Code CLI (a fork of Gemini CLI)", "excerpt": "I am actually doing now a good part of dev with Qwen3-Coder-Next on an M1 64GB with Qwen Code CLI (a fork of Gemini CLI). I very much like a) to have an idea how much tokens I use and b) be independent of VC financed token machines and c) I can use it on a plane/train Also I never have to wait in a queue, nor will I be told to wait for a few hours. And I get many answers in a second. I don't do full vibe coding with ", "context": "", "use_case": "", "switched_from": "", "story": "Ggml.ai joins Hugging Face to ensure the long-term progress of Local AI"}, {"id": 750358865, "item_id": 46897429, "url": "https://news.ycombinator.com/item?id=46897429", "by": "dust42", "date": "2026-02-05", "version": null, "stance": "recommend", "quote": "Together with Qwen Code (fork of Gemini) it is actually pretty capable.", "excerpt": "The brand new Qwen3-Coder-Next runs at 300Tok/s PP and 40Tok/s on M1 64GB with 4-bit MLX quant. Together with Qwen Code (fork of Gemini) it is actually pretty capable. Before that I used Qwen3-30B which is good enough for some quick javascript or Python, like 'add a new endpoint /api/foobar which does foobaz'. Also very decent for a quick summary of code. It is 530Tok/s PP and 50Tok/s TG. If you have it spit out lots", "context": "", "use_case": "local coding agent", "switched_from": "", "story": "Claude Code: connect to a local model when your quota runs out"}, {"id": 747061954, "item_id": 46691372, "url": "https://news.ycombinator.com/item?id=46691372", "by": "mx-Liu123", "date": "2026-01-20", "version": null, "stance": "recommend", "quote": "Qwen CLI is a great alternative if you want an OpenAI-compatible API with a generous free tier", "excerpt": "Author's Note: A few technical details for those looking to try AgentCommander: Why Gemini/Qwen CLI?: I chose these as backends because they offer robust directory isolation. I tried integrating Claude Code, but found it difficult to restrict its file-system reach. Qwen CLI is a great alternative if you want an OpenAI-compatible API with a generous free tier (2,000 requests/day). Environment: Ensure you have Python 3", "context": "", "use_case": "", "switched_from": "", "story": "Show HN: AgentCommander - workflow engine for evolutionary code optimization"}, {"id": 747061508, "item_id": 46691344, "url": "https://news.ycombinator.com/item?id=46691344", "by": "mx-Liu123", "date": "2026-01-20", "version": null, "stance": "recommend", "quote": "Built on top of Gemini/Qwen CLI to leverage their native tool-use capabilities", "excerpt": "I built AgentCommander to automate the manual \"trial-and-error\" loops in my PhD Physics/ML research. While tools like OpenEvolve (population evolution) and RD-Agent (Kaggle-style automation) exist, I found them difficult to customize for specific, multi-step research workflows. I needed a system that allowed granular control over the agent's decision process\u2014specifically, how it learns from errors and inherits code s", "context": "", "use_case": "", "switched_from": "", "story": "Show HN: AgentCommander - workflow engine for evolutionary code optimization"}, {"id": 734303729, "item_id": 45893983, "url": "https://news.ycombinator.com/item?id=45893983", "by": "bird0861", "date": "2025-11-11", "version": null, "stance": "recommend", "quote": "it does most of what I want", "excerpt": "Pretending 16 samples is authoritative is absolutely hilarious and wild, copium this pure could kill someone. Also working on a codebase you already know biases results in the first place -- they missed out on what has become a cornerstone of this stuff for AISWE people like me: repo tours; tree-sitter feeds the codebase to the LLM and I get to find all the stuff in the code I care about by either a single well forma", "context": "", "use_case": "Coding agent with swappable models", "switched_from": "", "story": "Baby Shoggoth Is Listening"}]}