Atlas

Models

← All models

Qwen3-Coder-Next

Alibaba · 2026-02-03 · 79.7B parameters

Qwen3-Coder-Next is Alibaba's open-weight, non-thinking coding model built on the Qwen3-Next hybrid Gated DeltaNet/Gated Attention mixture-of-experts architecture. Qwen reports 80B total and 3B active parameters at rounded precision, while the released checkpoint contains exactly 79,674,391,296 parameters. It is trained for agentic coding and long-horizon tool use, and supports a native 262,144-token context.

Benchmark scores

BenchmarkScore
AA-Briefcase Elo414
AA-Briefcase v1.1 Elo404
AA-LCR40.0
AA-LCR v1.147.0
AA-Omniscience Index-62.4
Artificial Analysis Agentic Index8.8
Artificial Analysis Coding Index36.2
Artificial Analysis Intelligence Index21.1
Artificial Analysis Intelligence Index v4.310.1
Artificial Analysis Intelligence Index v4.3.29.2
Artificial Analysis Omniscience Accuracy16.1
Artificial Analysis Omniscience Hallucination Rate93.7
Artificial Analysis Openness Index41.7
Artificial Analysis Progress Index v1227
AutomationBench-AA 1.0.61.1
BuseyBench SVG3.2
Capability138.2
CritPt (avg@5)0.0
GDP.pdf (Artificial Analysis, All-pass)2.6
GDP.pdf (Artificial Analysis, Mean Criteria)44.4
GDPval-AA v2664
GDPval-AA v2 (normalized)10.9
GDPval-AA v2.1522
Global MMLU Lite79.4
GPQA Diamond (Artificial Analysis avg@5)73.7
Humanity's Last Exam (Text-Only)10.1
Humanity's Last Exam Diamond4.3
IFBench (Artificial Analysis avg@5)35.2
ObviousBench43.1
Ruby LLM Benchmarks — Program Fixer56.7
Loading Atlas data…