What is Qwen3-30B-A3B FP8 good for?

Self-hosted assistant work on modest hardware, with switchable thinking and non-thinking modes. Apache 2.0 weights, a 30B mixture of experts with 3B active, on a single consumer GPU.

When is Qwen3-30B-A3B FP8 not a good fit?

Image, video or audio input: text only, and its native context limits long-document tasks.

What is the cheapest way to run Qwen3-30B-A3B FP8?

Hosted, unless you push serious volume. Alibaba Cloud charges $0.20 in / $0.80 out per 1M tokens. The cheapest rental that fits is 2x RTX 4070 at $144 a month, which costs the same as about 480M tokens a month on that API.

Can I self-host Qwen3-30B-A3B FP8?

Yes. The weights are Apache 2.0 licensed. At 4-bit it needs about 20 GB of GPU memory, which starts at roughly $144 a month on the cheapest rental that fits.

Open weights · Apache 2.0

Qwen3-30B-A3B FP8

FP8-quantized version of Qwen3-30B-A3B, a 30B MoE activating 3B parameters per token, with switchable thinking and non-thinking modes.

Jump to: Specs Pricing Self-hosting Similar models FAQ More from Alibaba Cloud

Visit website Compare vs other LLMs →

List price via Alibaba Cloud

Input: $0.20 /1M tokens
Output: $0.80 /1M tokens

Cheapest: $0.05 / $0.34 per 1M tokens via Cloudflare

3 providers

Key Specifications

Context window: 33K tokens
Parameters: 30.5B, 3.3B active
Inputs: Text

Capabilities Show details: Reasoning Thinks before it answers, always on or as a switchable mode. Function calling Connect to external tools, APIs, and systems. Structured output Return responses in structured formats like JSON.

Hosted API pricing

Input tokens M Output tokens M

Provider	Input /1M tokens	Output /1M tokens	Cost at 10M in + 2M out
Cloudflare Cheapest	$0.05	$0.34	$1.18	View
Lyceum	$0.10	$0.30	$1.60	View
Alibaba Cloud Creator	$0.20	$0.80	$3.60	View

Heads up: Base-tier, on-demand rates per 1M tokens; cached, batch and long-context tiers excluded. A provider may serve a shorter context or a quantized build than the creator's release. Verify before provisioning. More on how we price.

Estimated cost to self-host

Qwen3-30B-A3B FP8 needs about 20 GB of GPU memory at 4-bit with its full 32K context. The cheapest rental that fits is 2× RTX 4070 at about $144 a month.

That costs the same as roughly 480M tokens a month on Alibaba Cloud's API. Self-hosting is more expensive below that volume.

Single node ≤ 8

Precision	Memory	Cheapest fit (Full 32K context)	Cost /mo	Break-even vs API
4-bitINT4 / FP4	20 GB	2× RTX 4070	$144	480M tokens /mo
8-bitFP8 / INT8	36 GB	4× RTX 3060	$173	576M tokens /mo
16-bitFP16 / BF16	66 GB	8× RTX 3060	$346	1B tokens /mo

Estimates based on median on-demand rates for Nvidia GPUs. Memory is weights plus KV cache for one request, using FP8 KV cache where supported and FP16/BF16 otherwise. Break-even assumes a 5:1 input-to-output ratio. No guarantee of runtime support, usable performance, or that a matching quantized build exists. How we estimate costs.

Similarly priced models

The models nearest Qwen3-30B-A3B FP8 by blended rate, each at its own cheapest provider.

Model	Blended /1M	Input /1M	Output /1M	Context	Cutoff	vs Qwen3-30B-A3B FP8
GPT-OSS-120B OpenAI	$0.0633	$0.04	$0.18	131K	Jun 2024	−36%
Nemotron 3 Nano 30B A3B Nvidia	$0.075	$0.05	$0.20	262K	Jun 2025	−24%
Llama 3 8B Meta	$0.0833	$0.05	$0.25	8K	Mar 2023	−15%
Qwen3-Coder-30B-A3B Alibaba Cloud	$0.0917	$0.06	$0.25	262K		−7%
Llama 3.2 3B Meta	$0.0983	$0.05	$0.34	128K	Dec 2023	0%
Qwen3-30B-A3B FP8 This model Alibaba Cloud	$0.0983	$0.05	$0.34	33K
Ministral 3 3B Mistral	$0.10	$0.10	$0.10	256K		+2%
GLM-5.3-Flash Z.AI	$0.1083	$0.08	$0.25	1M		+10%
GPT-5 Nano OpenAI	$0.1083	$0.05	$0.40	400K	May 2024	+10%
Qwen3.5-35B-A3B Alibaba Cloud	$0.1267	$0.06	$0.46	262K		+29%
Qwen3.6-35B-A3B Alibaba Cloud	$0.1283	$0.07	$0.42	262K		+31%

Prices are USD per 1M tokens at each model's cheapest listed provider. Blended is the cost of 10M input plus 2M output tokens, spread over the 12M.

GPUhub

Fully Dedicated GPU Servers

No Virtualization. No Shared GPUs

Stable Performance for AI Training & Inference

Full Control for Production Workloads

Learn more

Frequently Asked Questions

What is Qwen3-30B-A3B FP8 good for?: Self-hosted assistant work on modest hardware, with switchable thinking and non-thinking modes. Apache 2.0 weights, a 30B mixture of experts with 3B active, on a single consumer GPU.
When is Qwen3-30B-A3B FP8 not a good fit?: Image, video or audio input: text only, and its native context limits long-document tasks.
What is the cheapest way to run Qwen3-30B-A3B FP8?: Hosted, unless you push serious volume. Alibaba Cloud charges $0.20 in / $0.80 out per 1M tokens. The cheapest rental that fits is 2x RTX 4070 at $144 a month, which costs the same as about 480M tokens a month on that API.
Can I self-host Qwen3-30B-A3B FP8?: Yes. The weights are Apache 2.0 licensed. At 4-bit it needs about 20 GB of GPU memory, which starts at roughly $144 a month on the cheapest rental that fits.

More from Alibaba Cloud

Model	Context	Input /1M	Output /1M
Qwen3-Coder-30B-A3B Open weights	262K	$0.06	$0.25
Qwen3.5-35B-A3B Open weights	262K	$0.06	$0.46
Qwen3.6-35B-A3B Open weights	262K	$0.07	$0.42
Qwen3-32B FP8 Open weights	33K	$0.10	$0.30
Qwen3.5-Flash	1M	$0.10	$0.40
Qwen3-235B-A22B Instruct (2507) Open weights	262K	$0.09	$0.58
Qwen3.5-27B Open weights	262K	$0.09	$0.69
Qwen3.8-Flash	1M	$0.15	$0.47