---
title: "Qwen 3 4B Benchmarks - RankedAGI"
description: "Detailed benchmark and metadata record for Qwen 3."
source: "https://rankedagi.com/models/qwen-3-4b-235b"
---

# Qwen 3

| Field | Value |
| --- | --- |
| Organization | Alibaba |
| License | Open Source |
| Size | 4B |
| Version | Latest |
| Released | 2025-04-29 |
| Context window | 32K |
| Last updated | 2025-06-08 10:44:06.341077+00 |

## Benchmarks

| Benchmark | Category | Value | Description | Source |
| --- | --- | --- | --- | --- |
| RankedAGI Coding | coding | 45.1% | RankedAGI Coding Score |  |
| SWEBench Verified | coding | 54.2% | Agentic Coding | https://www.swebench.com/index.html |
| RankedAGI Agentic | Agents | 37.0% | RankedAGI Agentic Score |  |
| Codeforces ELO Score | coding | 1671 | Codeforces ELO Score |  |
| RankedAGI Reasoning | reasoning | 39.9% | RankedAGI Reasoning Score |  |
| GPQA Diamond | reasoning | 55.9% | Generalized Prefix Question Answering Score (Reasoning) PhD Level Reasoning |  |
| AIME 2025 | math | 65.6% | AIME 2025 Competition Math |  |
| AIME 2024 | math | 73.8% | AIME 2024 Competition Math |  |
| Hughes Hallucination Evaluation | safety | 2.7% | Computed using Vectara's Hallucination Evaluation Model, also known as HHEM. This evaluates how often an LLM introduces hallucinations when summarizing a document. | https://github.com/vectara/hallucination-leaderboard |
| RankedAGI Math | math | 61.1% | RankedAGI Math Score |  |
| RankedAGI Overall | general | 44.7% | Overall RankedAGI score |  |

## Model Links

| Title | URL |
| --- | --- |
| Release | https://qwenlm.github.io/blog/qwen3/ |