Qwen3 Coder Next: benchmark results
Alibaba's efficiency-focused open Qwen3-Coder-Next agent MoE (80B total, 3B active). Provider: Alibaba. Released 2026-02-03. Access: Open.
Aggregate Rating 1642 ± 9, rank #326 of 1656 rated models, from 198 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| 6G-Bench - Intent and Policy Reasoning | 88.6 | Accuracy (%) on the intent and policy reasoning tasks (T1, T | 100 |
| Vector Eval - GPQA-D | 75.63 | Accuracy (%) | 100 |
| Vector Eval - InterCode-CTF | 85.9 | Accuracy (%) | 100 |
| Vector Eval - MATH | 97.82 | Accuracy (%) | 100 |
| 6G-Bench - Distributed Intelligence and Emerging Use Cases | 79.9 | Accuracy (%) on the distributed intelligence and emerging 6G | 96.3 |
| 6G-Bench - Network Slicing and Resource Management | 78.6 | Accuracy (%) on the network slicing and resource management | 96.3 |
| 6G-Bench | 81.8 | Accuracy (%) on the full set of 6G-Bench's 3,722 expert-vali | 95.2 |
| 6G-Bench - AI-Native Networking and Agentic Control | 83.8 | Accuracy (%) on the AI-native networking and agentic control | 90.7 |
| 6G-Bench - Trust, Security and SLA Awareness | 81.2 | Accuracy (%) on the trust, security and SLA awareness tasks | 90.4 |
| SecCodeBench | 62.76 | Total Score | 88.6 |
| Vector Eval - GSM8K | 95.83 | Accuracy (%) | 81.8 |
| Vector Eval - MMLU-Pro | 82.46 | Accuracy (%) | 81.8 |
Interactive version: theaggregate.ai/model?slug=qwen3-coder-next · How It Works · Data refreshed daily, snapshot 2026-10-20.