Qwen 3 Coder Plus — benchmark results

Alibaba's hosted agentic-coding API tier built on Qwen3-Coder 480B A35B MoE, with up to 1M context. Provider: Alibaba. Released 2025-09-23. Access: Open.

Unified ELO 1571 ± 30, rank #563 of 1776 rated models, from 14 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
WebCoderBench - Performance98.54Score (%)100
SnakeBench24.9TrueSkill Rating64
BenchTable51.3Total Score (%)54.9
MCPMark24.8Pass@1 (%)52.6
BacktestBench42.05Overall Accuracy (OA) (self-reported)45.5
AI Chess Leaderboard (Reasoning)639Elo45.2
WebCoderBench - Functionality Correctness54.95Score (%)30.8
Spider 2.0-Snow37.8Accuracy (%)30
ALE-Bench456.5Performance (Self-Refine x1) (self-reported)23.1
AssertLLM218.55Coverage Formal (self-reported)20
FrontierOR20Sol. quality (self-reported)16.7
WebCoderBench - Visual Experience46.27Score (%)15.4

Interactive version: theaggregate.ai/model?slug=qwen-3-coder-plus · How the rankings work · Data refreshed daily, snapshot 2026-07-22.