AIbase
EN
Home
AI Models Finder
Qwen3 30B A3B (Reasoning)
Q

Qwen3 30B A3B (Reasoning)

Go Compare
Qwen3-30B-A3B is a relatively small Mixture-of-Experts (MoE) model in Alibaba Cloud's Qwen3 series, with a total of 30.5 billion parameters and 3.3 billion active parameters. This model features a hybrid thinking/non-thinking mode, supports 119 languages, and has enhanced agent capabilities. Its goal is to outperform previous models such as QwQ-32B while significantly reducing the number of active parameters.
Intelligence(Medium)
Speed(Relatively Slow)
Input Supported Modalities
No
Is Reasoning Model
128,000
Context Window
128,000
Maximum Output Tokens
-
Knowledge Cutoff

Pricing

￥0.72 /M tokens
Input
￥2.16 /M tokens
Output
￥5.4 /M tokens
Blended Price

Quick Simple Comparison

Input

Output

Qwen2.5 Turbo
Qwen Turbo
Wan2.2

Basic Parameters

Qwen3 30B A3B (Reasoning)Technical Parameters
Parameter Count
30,500.0M
Context Length
128.00k tokens
Training Data Cutoff
Open Source Category
Open Weights (Permissive License)
Multimodal Support
Text Only
Throughput
122
Release Date
2025-04-28
Response Speed
89.55,036 tokens/s

Benchmark Scores

Below is the performance of Qwen3 30B A3B (Reasoning) in various standard benchmark tests. These tests evaluate the model's capabilities in different tasks and domains.
Intelligence Index
55.6
Large Language Model Intelligence Level
Coding Index
39.53
Indicator of AI model performance on coding tasks
Math Index
85.63
Capability indicator in solving mathematical problems, mathematical reasoning, or performing math-related tasks
MMLU Pro
77.7
Massive Multitask Multimodal Understanding - Testing understanding of text, images, audio, and video
GPQA
61.6
Graduate Physics Questions Assessment - Testing advanced physics knowledge with diamond science-level questions
HLE
6.6
The model's comprehensive average score on the Hugging Face Open LLM Leaderboard
LiveCodeBench
50.6
Specific evaluation focused on assessing large language models' ability in real-world code writing and solving programming competition problems
SciCode
28.5
The model's capability in code generation for scientific computing or specific scientific domains
HumanEval
-
Score achieved by the AI model on the specific HumanEval benchmark test set
Math 500 Score
95.9
Score on the first 500 larger, more well-known mathematical benchmark tests
AIME Score
75.3
An indicator measuring an AI model's ability to solve high-difficulty mathematical competition problems (specifically AIME level)
AIbase
Empowering the Future, Your AI Solution Knowledge Base
English简体中文繁體中文にほんご
© 2025AIbase