Qwen3-Coder-480B-A35B-Instruct at a glance
Qwen3-Coder-480B-A35B-Instruct supports 256K context natively (1M with extrapolation), was trained with large-scale code and agent reinforcement learning, and shipped with a terminal agent.
What changed with Qwen3-Coder-480B-A35B-Instruct
On July 22, 2025, the Qwen team announced Qwen3-Coder, its most agentic code model, starting with Qwen3-Coder-480B-A35B-Instruct: a mixture-of-experts model with 480 billion total and 35 billion active parameters. It natively handles 256K tokens of context and up to 1 million with YaRN extrapolation, aimed at repository-scale work.
The team said it was pretrained on 7.5 trillion tokens with a 70% code ratio and post-trained with execution-based reinforcement learning on real coding and multi-turn agent tasks. Qwen reported state-of-the-art results among open models on agentic coding, browser use and tool use, comparable to Claude Sonnet 4. It also open-sourced Qwen Code, a command-line agent adapted from Gemini CLI. The weights are Apache 2.0 licensed.
Qwen3-Coder-480B-A35B-Instruct in numbers
Values are copied from the linked source on the date shown. They are not IA Compa measurements.
- Context windowas of · Source for Context window: qwenlm.github.io (opens in a new tab)
- 256K tokens (native)
How it compares
Models that were current when this was published (released in the six months before it, newest version of each line), so these figures are from that time. Under each chart is the current best. (Models)
Current best (as of Nov 6, 2026): GPT-6 Cyber, 1.05M tokens
Where to compare Qwen3-Coder-480B-A35B-Instruct
Independent leaderboards and official resources for comparing this model yourself. Rankings change often; each site shows its own current data.
General leaderboards
Why Qwen3-Coder-480B-A35B-Instruct matters for businesses
For businesses, Qwen3-Coder gave development teams an open, self-hostable alternative for coding agents, useful when source code cannot be sent to a third-party API.
Sources for this story
- Qwen3-Coder: Agentic Coding in the World (opens in a new tab)Official
Qwen · qwenlm.github.io
- Qwen/Qwen3-Coder-480B-A35B-Instruct (opens in a new tab)Official
Hugging Face · huggingface.co
- Qwen3 Coder 480B model page (opens in a new tab)Benchmark
Artificial Analysis · artificialanalysis.ai
Last verified: . IA Compa summarizes public announcements for business readers. We are not affiliated with the companies mentioned. Always confirm details, pricing, and availability with the official source.
- #llm
- #alibaba
- #qwen
- #open-weights
- #coding
- #agents