---
title: "Qwen3 A3B Thinking 30B Benchmarks - RankedAGI"
description: "Detailed benchmark and metadata record for Qwen3 A3B Thinking."
source: "https://rankedagi.com/models/qwen3-30b-a3b-thinking"
---

# Qwen3 A3B Thinking

| Field | Value |
| --- | --- |
| Organization | Alibaba |
| License | Open Source |
| Size | 30B |
| Version | Latest |
| Released | 2025-07-30 |
| Context window | 1M |
| Last updated | 2026-04-20T06:27:11.080Z |

## Benchmarks

| Benchmark | Category | Value | Description | Source |
| --- | --- | --- | --- | --- |
| RankedAGI Coding | coding | 48.9% | RankedAGI Coding Score |  |
| RankedAGI Agentic | Agents | 38.7% | RankedAGI Agentic Score |  |
| LiveCodeBench v6 | coding | 66.0% | LiveCodeBench v6 - Evaluation of Large Language Models for Code |  |
| RankedAGI Reasoning | reasoning | 51.2% | RankedAGI Reasoning Score |  |
| GPQA Diamond | reasoning | 73.4% | Generalized Prefix Question Answering Score (Reasoning) PhD Level Reasoning |  |
| AIME 2025 | math | 85.0% | AIME 2025 Competition Math |  |
| MMLU Pro | knowledge | 80.9% | Graduate-level visual problem-solving |  |
| MMLU | knowledge | 91.4% | Image Reasoning |  |
| Instruction Following Evaluation | instruction-following | 88.9% | Instruction Following Evaluation |  |
| RankedAGI Math | math | 65.4% | RankedAGI Math Score |  |
| RankedAGI Overall | general | 48.6% | Overall RankedAGI score |  |
| GDPval-AA Elo | Agents | 660 | Office Tasks (Artificial Analysis) |  |

## Model Links

| Title | URL |
| --- | --- |
| Release | https://x.com/Alibaba_Qwen/status/1950570969036361799 |