Qwen3coder-30B is kinda dumb. It messes up tool calling and inserts manged XML tool calls into the context way too often. And it's **not** a great discriminator either with how much it's been RLHFed into automatic swebench discovery mode pretty much no matter what you ask it.
kv cache pressure from having two distinct conversations on a single llama.cpp running glm-4.5-air on my Ryzen AI Max+ 395 is immense. Down to 3 tokens/sec lmfaooooo. I'm going to have to revisit setting a default provider for each profile/agent.
December 6, 2025 at 7:03 PM
Qwen3coder-30B is kinda dumb. It messes up tool calling and inserts manged XML tool calls into the context way too often. And it's **not** a great discriminator either with how much it's been RLHFed into automatic swebench discovery mode pretty much no matter what you ask it.
Okay now I've got a qwen3coder-30B and a qwen3VL-30B running side by side on the AMD Ryzen AI Max+ 395. qwen3VL is crazy good. Like it is using playwright right now. This is fucking crazy dude. Local VLM lmfaoooooo. It's so on
I had 3 separarate qwen3-coder-30B instances running at the same time on ports 1234, 1235, and 1236 and connected three different zed agents to providers lmstudio, lmstudio-2, and lmstudio-3 [no actual lm studio involved] and it was slow as christmas but it worked.
November 24, 2025 at 1:07 AM
Okay now I've got a qwen3coder-30B and a qwen3VL-30B running side by side on the AMD Ryzen AI Max+ 395. qwen3VL is crazy good. Like it is using playwright right now. This is fucking crazy dude. Local VLM lmfaoooooo. It's so on
Code smarter with Qwen3 Coder in VSCode + Cline.
Top open-source coding model, right in your editor.
🎥 Quick guide: www.youtube.com/watch?v=paK...
#NebiusAIStudio #Nebius #Qwen3Coder #cline
Top open-source coding model, right in your editor.
🎥 Quick guide: www.youtube.com/watch?v=paK...
#NebiusAIStudio #Nebius #Qwen3Coder #cline
October 1, 2025 at 8:48 PM
Code smarter with Qwen3 Coder in VSCode + Cline.
Top open-source coding model, right in your editor.
🎥 Quick guide: www.youtube.com/watch?v=paK...
#NebiusAIStudio #Nebius #Qwen3Coder #cline
Top open-source coding model, right in your editor.
🎥 Quick guide: www.youtube.com/watch?v=paK...
#NebiusAIStudio #Nebius #Qwen3Coder #cline
Cerebras Challenges AI Coding Giants with High-Speed Access to Alibaba Qwen3-Coder Model
#AI #AICoding #DevTools #Cerebras #Alibaba #SoftwareDevelopment #Qwen3 #Qwen3Coder
winbuzzer.com/2025/08/02/c...
#AI #AICoding #DevTools #Cerebras #Alibaba #SoftwareDevelopment #Qwen3 #Qwen3Coder
winbuzzer.com/2025/08/02/c...
August 2, 2025 at 11:30 AM
Cerebras Challenges AI Coding Giants with High-Speed Access to Alibaba Qwen3-Coder Model
#AI #AICoding #DevTools #Cerebras #Alibaba #SoftwareDevelopment #Qwen3 #Qwen3Coder
winbuzzer.com/2025/08/02/c...
#AI #AICoding #DevTools #Cerebras #Alibaba #SoftwareDevelopment #Qwen3 #Qwen3Coder
winbuzzer.com/2025/08/02/c...
Qwen3-Coder の MLX版モデルの動作確認的な内容について、手順を記事に書いた！
（ #p5js のコードを書いてもらったもの ）
●M4 の Mac mini でローカルLLM： Qwen3-Coder-30B-A3B-Instruct（量子化された MLX版モデル）を MLX LM で試す - #Qiita
qiita.com/youtoy/items...
#Qwen3 #Qwen3Coder
（ #p5js のコードを書いてもらったもの ）
●M4 の Mac mini でローカルLLM： Qwen3-Coder-30B-A3B-Instruct（量子化された MLX版モデル）を MLX LM で試す - #Qiita
qiita.com/youtoy/items...
#Qwen3 #Qwen3Coder
M4 の Mac mini でローカルLLM： Qwen3-Coder-30B-A3B-Instruct（量子化された MLX版モデル）を MLX LM で試す - Qiita
はじめに 公式からは以下でポストされていた Qwen/Qwen3-Coder-30B-A3B-Instruct について、量子化された MLX版モデルを試した話です。 使ったモデルとお試し用環境 今回用いた、MLX版で量子化されたものの 1つは以下になります。 ●...
qiita.com
August 2, 2025 at 11:09 AM
Qwen3-Coder の MLX版モデルの動作確認的な内容について、手順を記事に書いた！
（ #p5js のコードを書いてもらったもの ）
●M4 の Mac mini でローカルLLM： Qwen3-Coder-30B-A3B-Instruct（量子化された MLX版モデル）を MLX LM で試す - #Qiita
qiita.com/youtoy/items...
#Qwen3 #Qwen3Coder
（ #p5js のコードを書いてもらったもの ）
●M4 の Mac mini でローカルLLM： Qwen3-Coder-30B-A3B-Instruct（量子化された MLX版モデル）を MLX LM で試す - #Qiita
qiita.com/youtoy/items...
#Qwen3 #Qwen3Coder
#Cerebras Code launches with #Qwen3Coder at 2,000 tokens/second 🚀
#ai #coding #subscription
💰 Two pricing tiers: Code Pro ($50/month) and Code Max ($200/month) for fast #AI code generation
🔥 Qwen3-Coder 480B model delivers performance comparable to #Claude Sonnet 4 and #GPT4
🧵 👇 #AI #coding
#ai #coding #subscription
💰 Two pricing tiers: Code Pro ($50/month) and Code Max ($200/month) for fast #AI code generation
🔥 Qwen3-Coder 480B model delivers performance comparable to #Claude Sonnet 4 and #GPT4
🧵 👇 #AI #coding
August 1, 2025 at 7:58 PM
#Cerebras Code launches with #Qwen3Coder at 2,000 tokens/second 🚀
#ai #coding #subscription
💰 Two pricing tiers: Code Pro ($50/month) and Code Max ($200/month) for fast #AI code generation
🔥 Qwen3-Coder 480B model delivers performance comparable to #Claude Sonnet 4 and #GPT4
🧵 👇 #AI #coding
#ai #coding #subscription
💰 Two pricing tiers: Code Pro ($50/month) and Code Max ($200/month) for fast #AI code generation
🔥 Qwen3-Coder 480B model delivers performance comparable to #Claude Sonnet 4 and #GPT4
🧵 👇 #AI #coding
Alibaba’s AI Coder Raises Eyebrows
Alibaba’s Qwen3-Coder is a powerful new AI dev tool built on a 35B-parameter model. But Western experts are raising red flags over security and influence. Innovation or intrusion?
#AI #Alibaba #Qwen3Coder #AICoding
www.efficiencyai.co.uk/alibabas-new...
Alibaba’s Qwen3-Coder is a powerful new AI dev tool built on a 35B-parameter model. But Western experts are raising red flags over security and influence. Innovation or intrusion?
#AI #Alibaba #Qwen3Coder #AICoding
www.efficiencyai.co.uk/alibabas-new...
Alibaba’s New AI Coding Tool Qwen3-Coder Sparks Western Concerns - Efficiency AI Transformation
Alibaba has introduced Qwen3-Coder, a sophisticated AI tool designed to assist with complex software development. This innovative tool is built on a 35-billion-parameter Mixture of Experts (MoE) open-...
www.efficiencyai.co.uk
July 31, 2025 at 11:51 AM
Alibaba’s AI Coder Raises Eyebrows
Alibaba’s Qwen3-Coder is a powerful new AI dev tool built on a 35B-parameter model. But Western experts are raising red flags over security and influence. Innovation or intrusion?
#AI #Alibaba #Qwen3Coder #AICoding
www.efficiencyai.co.uk/alibabas-new...
Alibaba’s Qwen3-Coder is a powerful new AI dev tool built on a 35B-parameter model. But Western experts are raising red flags over security and influence. Innovation or intrusion?
#AI #Alibaba #Qwen3Coder #AICoding
www.efficiencyai.co.uk/alibabas-new...
AIクリエーターの道 ニュース: 開発効率爆上げ！アリババの「Qwen3-Coder」が登場。複雑なコーディングも自動化！ #Qwen3Coder #AIコーディング #アリババ
詳しくはこちら↓↓↓
gamefi.co.jp/2025/07/24/a...
詳しくはこちら↓↓↓
gamefi.co.jp/2025/07/24/a...
アリババ、新AIコーディングモデル「Qwen3-Coder」発表 | AI News
アリババが革新的なAIコーディングモデル「Qwen3-Coder」をリリース！開発を加速する最新技術を紹介。
gamefi.co.jp
July 26, 2025 at 9:00 PM
AIクリエーターの道 ニュース: 開発効率爆上げ！アリババの「Qwen3-Coder」が登場。複雑なコーディングも自動化！ #Qwen3Coder #AIコーディング #アリババ
詳しくはこちら↓↓↓
gamefi.co.jp/2025/07/24/a...
詳しくはこちら↓↓↓
gamefi.co.jp/2025/07/24/a...
🔷 Qwen 3 Coder support is now live on Forge! We ran head‑to‑head tests versus two #opensource models, Kimi K2 on a real‑world 38K‑line Rust & 12K‑line React codebase. See how they compare.
#AI #CodingAgent #Qwen3Coder #KimiK2 #Rust #React
#AI #CodingAgent #Qwen3Coder #KimiK2 #Rust #React
July 24, 2025 at 7:20 AM
🔷 Qwen 3 Coder support is now live on Forge! We ran head‑to‑head tests versus two #opensource models, Kimi K2 on a real‑world 38K‑line Rust & 12K‑line React codebase. See how they compare.
#AI #CodingAgent #Qwen3Coder #KimiK2 #Rust #React
#AI #CodingAgent #Qwen3Coder #KimiK2 #Rust #React
AIMindUpdate News!
Unleash your coding potential! Alibaba's new Qwen3-Coder is here to revolutionize your development workflow. #AIcoding #Qwen3Coder #AlibabaAI
Click here↓↓↓
aimindupdate.com/2025/07/24/a...
Unleash your coding potential! Alibaba's new Qwen3-Coder is here to revolutionize your development workflow. #AIcoding #Qwen3Coder #AlibabaAI
Click here↓↓↓
aimindupdate.com/2025/07/24/a...
Alibaba's Qwen3-Coder: Advanced AI for Coding | AI News
Discover Alibaba's Qwen3-Coder: the most powerful AI coding model yet, optimized for complex tasks.
aimindupdate.com
July 24, 2025 at 5:00 AM
AIMindUpdate News!
Unleash your coding potential! Alibaba's new Qwen3-Coder is here to revolutionize your development workflow. #AIcoding #Qwen3Coder #AlibabaAI
Click here↓↓↓
aimindupdate.com/2025/07/24/a...
Unleash your coding potential! Alibaba's new Qwen3-Coder is here to revolutionize your development workflow. #AIcoding #Qwen3Coder #AlibabaAI
Click here↓↓↓
aimindupdate.com/2025/07/24/a...
💻 Meet #Qwen3Coder — Alibaba’s open-source code model for 350+ languages.
#AI #Qwen3Coder #AlibabaCloud #CodeGeneration #OpenSourceAI #DeveloperTools #LLM
Read more:
aiadoptionagency.com/qwen3-coder-...
#AI #Qwen3Coder #AlibabaCloud #CodeGeneration #OpenSourceAI #DeveloperTools #LLM
Read more:
aiadoptionagency.com/qwen3-coder-...
Qwen3-Coder: Alibaba’s Open-Source Powerhouse for Code Generation and Software Development - Ai Adoption Agency
Qwen3-Coder is Alibaba Cloud’s flagship open-source code generation model within the Qwen3 series, created to write, debug, and optimize software using natural language prompts. Released in July 2025 ...
https://aiadoptionagency.com/qwen3-coder-alibabas-open-source-powerhouse-for-code-generation-and-software-development/"
July 23, 2025 at 8:48 PM
💻 Meet #Qwen3Coder — Alibaba’s open-source code model for 350+ languages.
#AI #Qwen3Coder #AlibabaCloud #CodeGeneration #OpenSourceAI #DeveloperTools #LLM
Read more:
aiadoptionagency.com/qwen3-coder-...
#AI #Qwen3Coder #AlibabaCloud #CodeGeneration #OpenSourceAI #DeveloperTools #LLM
Read more:
aiadoptionagency.com/qwen3-coder-...
🚀 A corrida global pela IA esquenta!
O Alibaba acaba de lançar o Qwen3-Coder, seu modelo de codificação mais avançado e open source.
A China mira o topo, desafiando o GitHub Copilot e OpenAI na arena da IA para desenvolvedores.
Código, competição e inovação. 🇨🇳💻🤖 #IA #Alibaba #Qwen3Coder #OpenSource
O Alibaba acaba de lançar o Qwen3-Coder, seu modelo de codificação mais avançado e open source.
A China mira o topo, desafiando o GitHub Copilot e OpenAI na arena da IA para desenvolvedores.
Código, competição e inovação. 🇨🇳💻🤖 #IA #Alibaba #Qwen3Coder #OpenSource
July 23, 2025 at 1:58 PM
🚀 A corrida global pela IA esquenta!
O Alibaba acaba de lançar o Qwen3-Coder, seu modelo de codificação mais avançado e open source.
A China mira o topo, desafiando o GitHub Copilot e OpenAI na arena da IA para desenvolvedores.
Código, competição e inovação. 🇨🇳💻🤖 #IA #Alibaba #Qwen3Coder #OpenSource
O Alibaba acaba de lançar o Qwen3-Coder, seu modelo de codificação mais avançado e open source.
A China mira o topo, desafiando o GitHub Copilot e OpenAI na arena da IA para desenvolvedores.
Código, competição e inovação. 🇨🇳💻🤖 #IA #Alibaba #Qwen3Coder #OpenSource
Qwen3-Coder's release sparks Hacker News debate! Community explores its coding capabilities, local deployment via quantization, and comparisons to Claude Code. Big questions: model size vs. performance, and how AI reshapes software engineering workflows. #Qwen3Coder 1/6
July 23, 2025 at 1:00 PM
Qwen3-Coder's release sparks Hacker News debate! Community explores its coding capabilities, local deployment via quantization, and comparisons to Claude Code. Big questions: model size vs. performance, and how AI reshapes software engineering workflows. #Qwen3Coder 1/6
#Qwen3Coder: Most Agentic Code Model Released 🤖
🎯 480B-parameter #MixtureOfExperts #LLM with 35B active parameters achieving #SOTA performance in agentic #coding
📏 Native 256K context support, extendable to 1M
tokens with #YaRN for repo-scale operations
🧵👇#AI
qwenlm.github.io/blog/qwen3-...
🎯 480B-parameter #MixtureOfExperts #LLM with 35B active parameters achieving #SOTA performance in agentic #coding
📏 Native 256K context support, extendable to 1M
tokens with #YaRN for repo-scale operations
🧵👇#AI
qwenlm.github.io/blog/qwen3-...
Qwen3-Coder: Agentic Coding in the World
GITHUB HUGGING FACE MODELSCOPE DISCORD
Today, we’re announcing Qwen3-Coder, our most agentic code model to date. Qwen3-Coder is available in multiple sizes, but we’re excited to introduce its most powerful variant first: Qwen3-Coder-480B-A35B-Instruct — a 480B-parameter Mixture-of-Experts model with 35B active parameters which supports the context length of 256K tokens natively and 1M tokens with extrapolation methods, offering exceptional performance in both coding and agentic tasks. Qwen3-Coder-480B-A35B-Instruct sets new state-of-the-art results among open models on Agentic Coding, Agentic Browser-Use, and Agentic Tool-Use, comparable to Claude Sonnet 4.
qwenlm.github.io
July 23, 2025 at 11:45 AM
#Qwen3Coder: Most Agentic Code Model Released 🤖
🎯 480B-parameter #MixtureOfExperts #LLM with 35B active parameters achieving #SOTA performance in agentic #coding
📏 Native 256K context support, extendable to 1M
tokens with #YaRN for repo-scale operations
🧵👇#AI
qwenlm.github.io/blog/qwen3-...
🎯 480B-parameter #MixtureOfExperts #LLM with 35B active parameters achieving #SOTA performance in agentic #coding
📏 Native 256K context support, extendable to 1M
tokens with #YaRN for repo-scale operations
🧵👇#AI
qwenlm.github.io/blog/qwen3-...
Alibaba Targets Agentic AI Crown with Qwen3-Coder Release
#AI #Alibaba #Qwen #Qwen3Coder #OpenSourceAI #AIAgents #DeveloperTools #AgenticAI
winbuzzer.com/2025/07/23/a...
#AI #Alibaba #Qwen #Qwen3Coder #OpenSourceAI #AIAgents #DeveloperTools #AgenticAI
winbuzzer.com/2025/07/23/a...
July 23, 2025 at 8:31 AM
Alibaba Targets Agentic AI Crown with Qwen3-Coder Release
#AI #Alibaba #Qwen #Qwen3Coder #OpenSourceAI #AIAgents #DeveloperTools #AgenticAI
winbuzzer.com/2025/07/23/a...
#AI #Alibaba #Qwen #Qwen3Coder #OpenSourceAI #AIAgents #DeveloperTools #AgenticAI
winbuzzer.com/2025/07/23/a...
