Qwen3 Coder Next is an open-weight coding model built on Qwen3-Next-80B-A3B-Base (hybrid attention + MoE). It is agentically trained at scale on executable tasks and environment interaction, delivering strong coding and tool-use performance at lower inference cost. Native 256K context.
Added Dec 8, 2025
Model weightsContext Window
262.1K
Max Output
65.5K
Input Price (Auto)
$0.21/1M
Output Price (Auto)
$1.57/1M
Cache Read (Auto)
$0.11/1M
Capabilities
Benchmarks
Benchmarks
Performance metrics and benchmarks
Sourced from Artificial Analysis.
Intelligence Index
21.2
Coding Index
36.2
Agentic Index
8.9
Reasoning
GPQA Diamond
Graduate-level scientific reasoning
73.7%
Better than 62% of models compared
HLE
Humanity's Last Exam
9.3%
Better than 63% of models compared
IFBench
Instruction-following benchmark
35.2%
Better than 28% of models compared
T²-Bench Telecom
Conversational AI agents in dual-control scenarios
79.5%
Better than 70% of models compared
AA-LCR
Long context reasoning evaluation
40.0%
Better than 54% of models compared
GDPval-AA
Economically valuable tasks
10.8%
CritPt
Research-level physics reasoning
0.0%
Coding
SciCode
Python programming for scientific computing
32.3%
Better than 51% of models compared
Terminal-Bench Hard
Agentic coding and terminal use
18.2%
Better than 59% of models compared
Knowledge
AA-Omniscience Accuracy
Proportion of correctly answered questions
16.2%
AA-Omniscience Hallucination Rate
Rate of incorrect answers among non-correct responses
93.7%
Last updated Jun 28, 2026
Artificial AnalysisProviders
Choose explicit providers for this model. Auto routing remains available as the default option.
Loading provider options…
Related text models
Compare Qwen3 Coder Next with similar models from the same provider or model family.
Qwen 3 Coder 480B
qwen/qwen3-coderQwen 3 Coder 480B, a 480 billion total parameter model with 35B active, and 160 total experts with 8 active. Performs similar to Claude 4 Sonnet in coding benchmarks, but does so at a much lower price.
Qwen3 Coder Flash
qwen/qwen3-coder-flashA speed‑optimized and budget‑friendly sibling to Coder Plus. Excellent at code generation and agentic workflows (tool use, environment interaction) with solid general‑purpose ability.
Qwen3 Coder Plus
qwen/qwen3-coder-plusAlibaba’s proprietary upgrade to the open‑weights Qwen3 Coder 480B A35B. A coding‑first agent model with strong tool use and environment control for autonomous programming, while remaining capable at general tasks.
Qwen3 Next 80B A3B (Instruct)
Qwen/Qwen3-Next-80B-A3B-InstructBased on the new Qwen3‑Next architecture (hybrid attention, highly sparse MoE, training‑stability optimizations, and multi‑token prediction), the Qwen3‑Next‑80B‑A3B‑Instruct model delivers extreme efficiency with only 3B active parameters per pass. It performs comparably to Qwen3‑235B‑A22B‑Instruct‑2507 and shows clear advantages on ultra‑long context tasks (up to 256K tokens).
Qwen3 Next 80B A3B (Thinking)
qwen/qwen3-next-80b-a3b-thinkingQwen3 Next 80B A3B (Thinking)
Qwen3.6 35B A3B Thinking
Qwen/Qwen3.6-35B-A3B:thinkingQwen3.6 35B A3B is a native vision-language MoE model with hybrid attention. Compared to Qwen3.5 35B A3B, Alibaba reports stronger agentic coding, mathematical and code reasoning, and better spatial understanding (including object localization and detection).