Qwen3-Coder-Next
Alibaba's 80B-A3B open coding specialist on the Qwen3-Next architecture.
About
Qwen3-Coder-Next is Alibaba's open-weight coding specialist released in February 2026, built on the Qwen3-Next hybrid linear-attention architecture. It is a mixture-of-experts model with 80B total parameters and only 3B active per token, released under Apache 2.0 with a 256K-token context window.
It is designed for agentic coding: multi-step tool use, long-horizon task execution, and repository-scale reasoning. Vendor-reported figures put it at 70.6 percent on SWE-bench Verified with SWE-Agent scaffolding, and community testing has highlighted its performance-per-cost story against hosted coding models.
The per-token compute tracks the small active set, but the full expert pool still needs roughly 43GB of memory at 4-bit, so self-hosting wants a 64GB-class machine. It integrates with Ollama, LM Studio, llama.cpp, and coding agents like Claude Code, Cline, and Trae.
Specs
Context window
Pricing
Free quota
API access
Weights
Released
Benchmarks
Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.
Use cases
Trust score — 9/10
Scored 9/10 (verified tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 104; Trending score 86/100 from researcher assessment.
More in the qwen3 family
- Qwen3-32B-AWQ
- Qwen3-Next-80B-A3B
- Qwen3-235B-A22B
- Qwen3-235B-A22B-2507
- Qwen3-4B-Instruct-2507
- Qwen3-8B
- Qwen3.5
- Qwen3-235B-A22B-Instruct-2507
- Qwen3 235B
- Qwen3-235B-A22B-Thinking-2507
- Qwen3-Omni-Flash
- Qwen3.6-27B
- Qwen3-32B
- Qwen3-Max
- Qwen3.8-Flash-Next
- Qwen3-Coder-Plus
- Qwen3
- Qwen3-0.6B
- Qwen3-4B
- Qwen3-4B-2507
- Qwen3-14B
- Qwen3-VL-30B-A3B
- Qwen3-1.7B
- Qwen3-Omni-30B-A3B