modelgrep
Q

Qwen: Qwen3 30B A3B Instruct 2507

qwen/qwen3-30b-a3b-instruct-2507

143rd smartest of 178Cheaper than 95% of paidToolsJSON
Use via OpenRouter ↗
Intelligence
15.0
143rd of 178
Design Elo
Speed
72
106th fastest
Latency
221ms
first token
Input price
$0.048
14th cheapest
Context
131K
32K max out

How it compares

Smarter than20%
of all ranked models
Faster than64%
of all ranked models
Cheaper than95%
of all ranked models

Overview

Qwen3-30B-A3B-Instruct-2507 is a 30.5B-parameter mixture-of-experts language model from Qwen, with 3.3B active parameters per inference. It operates in non-thinking mode and is designed for high-quality instruction following, multilingual understanding, and...

Benchmarks

independent · via OpenRouter
Artificial Analysis20th percentile
Intelligence Index
15.0
Coding Index
14.2
Agentic Index
7.1
GPQA Diamond
66%
Humanity's Last Exam
7%
SciCode
30%
Tau²-Bench (agentic)
10%

Providers & pricing (6)

ProviderIn $/MOut $/MUptime
StreamLake$0.048$0.19399.9%
SiliconFlowfp8$0.090$0.30069.7%
Nebiusfp8$0.100$0.300100%
AtlasCloudfp8$0.100$0.30099.8%
WandBbf16$0.100$0.300100%
Alibaba$0.130$0.52099.9%

Specifications

Context window131K
Max output32K
Knowledge cutoffJun 2025
Input modalitiestext
Output modalitiestext
Prompt caching
Cache read price
ModeratedNo

Qwen3 30B A3B Instruct 2507 FAQ

How much does Qwen3 30B A3B Instruct 2507 cost?

Qwen3 30B A3B Instruct 2507 costs $0.048 per million input tokens and $0.193 per million output tokens via OpenRouter, making it 14th cheapest of 298 paid models.

How smart is Qwen3 30B A3B Instruct 2507?

Qwen3 30B A3B Instruct 2507 scores 15.0 on the Artificial Analysis Intelligence Index, ranking 143rd of 178 benchmarked models, with a GPQA Diamond score of 66%.

How fast is Qwen3 30B A3B Instruct 2507?

Qwen3 30B A3B Instruct 2507 generates around 72 tokens per second with 221ms time-to-first-token (p50), the 106th fastest tracked model.

What is Qwen3 30B A3B Instruct 2507's context window?

Qwen3 30B A3B Instruct 2507 supports a 131K-token context window and can output up to 32K tokens. It accepts text input.

Compare head-to-head