Alibaba (Qwen)

Qwen3-30B-A3B

Generally availableOpen weights41K contextReasoning

Summary

Qwen3-30B-A3B is an open-weight language model from Alibaba (Qwen), released on 28 Apr 2025. It accepts text and generates text, with a 41K-token context window and up to 16K output tokens per response. As of 18 Nov 2026, the lowest listed API price is $0.12 per million input tokens and $0.50 per million output tokens (DeepInfra), across 4 providers we track.

Qwen instruction model for multilingual chat, reasoning, and tool use

Last verified 18 Nov 2026. Every figure on this page links to its source.

Specifications

DeveloperAlibaba (Qwen)
Release date28 Apr 2025 [models.dev]
StatusGenerally available [models.dev]
WeightsOpen weights [Hugging Face]
Licenseapache-2.0 [Hugging Face]
Parameters30.5B [Hugging Face]
Context window41K (40,960 tokens) [models.dev]
Max output16K tokens [models.dev]
Knowledge cutoffJun 2025 [models.dev]
InputText [models.dev]
OutputText [models.dev]
Reasoning modeYes [models.dev]
Tool callingYes [models.dev]
Structured outputYes [models.dev]
Hugging FaceQwen/Qwen3-30B-A3B [Hugging Face]

Qwen3-30B-A3B API pricing by provider

USD per million tokens. Sorted by input price.

ProviderInputOutputCached inputContextSource
DeepInfra
Qwen/Qwen3-30B-A3B
$0.12$0.50–41Kmodels.dev, 18 Nov 2026
Hugging Face Inference Providers
Qwen/Qwen3-30B-A3B
$0.12$0.50–41Kmodels.dev, 18 Nov 2026
OpenRouter
qwen/qwen3-30b-a3b
$0.12$0.50–41KOpenRouter, 18 Nov 2026
Vercel AI Gateway
alibaba/qwen-3-30b
$0.12$0.50–41Kmodels.dev, 18 Nov 2026

Running Qwen3-30B-A3B locally

Estimated memory, from the parameter count and the architecture in config.json on Hugging Face. Includes about 0.8 GB of runtime overhead.

QuantizationWeightsTotal, 8K contextTotal, 32K context
Q4_K_M (4-bit)19 GB20 GB23 GB
Q8_0 (8-bit)32 GB34 GB36 GB
FP16 / BF16 (16-bit)61 GB63 GB65 GB

At 4-bit with an 8K conversation, Qwen3-30B-A3B needs about 20 GB, so it fits on a graphics card with 24 GB or more, such as the NVIDIA GeForce RTX 3090 24 GB. Check your own hardware, or see Best GPU for local LLMs on our sister site AIPCs.

Frequently asked questions

What is the context window of Qwen3-30B-A3B?

Qwen3-30B-A3B has a context window of 40,960 tokens (41K) and can generate up to 16,384 tokens in a single response, according to models.dev.

How much does the Qwen3-30B-A3B API cost?

Qwen3-30B-A3B is listed from $0.12 per million input tokens and $0.50 per million output tokens (DeepInfra). Prices last verified 18 Nov 2026.

Is Qwen3-30B-A3B open source?

Qwen3-30B-A3B is an open-weight model: its weights are published on Hugging Face as Qwen/Qwen3-30B-A3B under the apache-2.0 license. Check the license terms before commercial use.

When was Qwen3-30B-A3B released?

Alibaba (Qwen) released Qwen3-30B-A3B on 28 Apr 2025, according to models.dev.

Which providers offer Qwen3-30B-A3B?

We track Qwen3-30B-A3B on 4 providers: DeepInfra, Hugging Face Inference Providers, OpenRouter, Vercel AI Gateway.

How much VRAM does Qwen3-30B-A3B need?

About 20 GB at 4-bit (Q4_K_M) with an 8K-token conversation, and about 34 GB at 8-bit. The 4-bit model fits on a graphics card with 24 GB or more, such as the NVIDIA GeForce RTX 3090 24 GB. These are estimates from the model's published parameter count and architecture.

Similar models from other labs

Comparable on output type, weights, release date, price and context window. Prices are the lowest listed, USD per million tokens.

ModelLabReleasedContextInput $/1MOutput $/1MWeights
Voxtral SmallMistral AI15 Jul 202533K$0.10$0.40Open
Voxtral Small 24B 2507Mistral AI15 Jul 202533K$0.10$0.30Open
Skyfall 36B V2Thedrummer10 Mar 202533K$0.55$0.80Open
Reka Flash 3Rekaai10 Mar 202566K$0.10$0.20Open
Llama-4-Scout-17B-16E-Instruct-FP8Meta5 Apr 2025128K$0.10$0.30Open
DeepSeek-V3DeepSeek26 Dec 202464K$0.32$0.89Open

Change history

  • Knowledge cutoff changed from Apr 2025 to Jun 2025
  • Name changed from Qwen3 30B A3B to Qwen3-30B-A3B
  • Name changed from Qwen3-30B-A3B to Qwen3 30B A3B
  • Context window changed from 128K to 41K
  • Max output changed from 8K to 16K
  • Reasoning mode changed from no to yes
  • Max output changed from 16K to 8K
  • Knowledge cutoff changed from Jun 2025 to Apr 2025
  • Reasoning mode changed from yes to no
  • Context window changed from 41K to 128K
  • Name changed from Qwen3 30B A3B to Qwen3-30B-A3B
  • Name changed from Qwen3-30B-A3B to Qwen3 30B A3B
  • Max output changed from 8K to 16K
  • Knowledge cutoff changed from Apr 2025 to Jun 2025
  • Reasoning mode changed from no to yes
  • Context window changed from 128K to 41K
  • Max output changed from 41K to 8K
  • Knowledge cutoff changed from Jun 2025 to Apr 2025
  • Name changed from Qwen3 30B A3B to Qwen3-30B-A3B
  • Context window changed from 41K to 128K
  • Max output changed from 16K to 41K
  • Reasoning mode changed from yes to no
  • Context window changed from 41K to 41K
  • Name changed from Qwen3-30B-A3B to Qwen3 30B A3B
  • Knowledge cutoff changed from Apr 2025 to Jun 2025

Sources