Skip to main content
Language modelActive

Qwen 3 30B A3B Instruct 2507

Alibaba

Released
July 28, 2025
Data date
September 8, 2026
Predecessor
Qwen 3 30B A3B
Speed
150.5 tokens/s
Output throughput
Price
$0.09 · $0.3
Input · Output / 1M tokens
Input
Text
Output
Text

Qwen 3 30B A3B Instruct 2507 is the compact direct chat configuration in the Qwen3 MoE line. It has 30.5 billion total parameters and activates 3.3 billion. The checkpoint runs in non-thinking mode and provides 262,144 context tokens. Qwen also documents extension to 1,010,000 tokens.

The Apache-2.0 weights use the non-thinking configuration without visible reasoning. For visible reasoning, use Qwen 3 30B A3B Thinking 2507, the other configuration in this class. The pretrained Qwen 3 30B A3B shares the parameter class but not the same chat tuning.

Specifications and access

SpecificationValue and source
Weights
Open weightsSource
Access
API, LocalSource
Parameters
30.5BSource
Active parameters
3.3BSource
Context window
262,144 tokensSource
Weights repository
Qwen/Qwen3-30B-A3B-Instruct-2507Source
License
Apache-2.0Source
Architecture
Qwen3 causal language model with sparse MoE and 8 activated expertsSource
Input
TextSource
Output
TextSource
Reasoning mode
non-thinking onlySource
Tool calling
YesSource
Structured output
YesSource
Native context window
262,144 tokensSource
Maximum context window
1,010,000 tokensSource
Parameters
30.5 BSource
Active parameters
3.3 BSource
Languages
119 languages and dialectsSource
Repository created
2025-07-28 (Repository creation date is a precise publication proxy, but may differ from the public announcement date or an individual checkpoint's first availability.)Source
Training tokens
36 trillionSource