Qwen3-30B-A3B

reasoning

Qwen3-30B-A3B is the first Mixture-of-Experts (MoE) model in the series, featuring a total of 30 billion parameters, with only 3 billion activated per token. The model is built on an architecture with 48 layers, 32 attention heads, and supports a context window of 128K tokens.

The model employs an innovative Mixture of Experts (MoE) architecture with a total of 128 experts, out of which only 8 are activated per token. This allows achieving inference efficiency comparable to compact 3B-parameter models, while the generation quality approaches that of dense 30B-parameter models. The key advantage of the MoE approach is its optimal balance between high performance and energy efficiency. The system dynamically selects the most suitable experts for each specific task, delivering excellent processing quality at significantly lower computational costs compared to dense models of similar capability.

Qwen3-30B-A3B is well-suited for high-load systems requiring a balance between quality and performance: cloud API services, enterprise chatbots, content automation systems, multi-user AI platforms. The model is ideal for companies that need the quality of large models but operate under limited computational budgets.


Announce Date: 29.04.2025
Parameters: 30.5B
Experts: 128
Activated: 3.3B
Context: 131K
Attention Type: Full or Sliding Window Attention
VRAM requirements: 26.2 GB using 4 bits quantization
Developer: Alibaba
Transformers Version: 4.51.0
Ollama Version: 0.6.6
License: Apache 2.0

Public endpoint

Use our pre-built public endpoints to test inference and explore Qwen3-30B-A3B capabilities.
Model Name Context Type GPU TPS Status Link
There are no public endpoints for this model yet.

Private server

Rent your own physically dedicated instance with hourly or long-term monthly billing.

We recommend deploying private instances in the following scenarios:

  • maximize endpoint performance,
  • enable full context for long sequences,
  • ensure top-tier security for data processing in an isolated, dedicated environment,
  • use custom weights, such as fine-tuned models or LoRA adapters.

Recommended configurations for hosting Qwen3-30B-A3B

Prices:
Name vCPU RAM, MB Disk, GB GPU Price, hour
teslaa2-2.16.32.160 16 32768 160 2 $0.57 Launch
teslat4-2.16.32.160 16 32768 160 2 $0.80 Launch
teslaa10-2.16.64.160 16 65536 160 2 $0.93 Launch
rtx2080ti-3.16.64.160 16 65536 160 3 $0.95 Launch
teslav100-1.12.64.160 12 65536 160 1 $1.20 Launch
rtx3080-3.16.64.160 16 65536 160 3 $1.43 Launch
rtx5090-1.16.64.160 16 65536 160 1 $1.59 Launch
rtx3090-2.16.64.160 16 65536 160 2 $1.67 Launch
rtx4090-2.16.64.160 16 65536 160 2 $2.19 Launch
teslaa100-1.16.64.160 16 65536 160 1 $2.58 Launch
teslah100-1.16.64.160 16 65536 160 1 $5.11 Launch
Prices:
Name vCPU RAM, MB Disk, GB GPU Price, hour
teslaa10-2.16.64.160 16 65536 160 2 $0.93 Launch
teslat4-4.16.64.160 16 65536 160 4 $1.48 Launch
rtx3090-2.16.64.160 16 65536 160 2 $1.67 Launch
rtx4090-2.16.64.160 16 65536 160 2 $2.19 Launch
teslav100-2.16.64.240 16 65535 240 2 $2.22 Launch
teslaa100-1.16.64.160 16 65536 160 1 $2.58 Launch
rtx5090-2.16.64.160 16 65536 160 2 $2.93 Launch
teslah100-1.16.64.160 16 65536 160 1 $5.11 Launch
Prices:
Name vCPU RAM, MB Disk, GB GPU Price, hour
teslaa10-4.16.128.160 16 131072 160 4 $1.75 Launch
teslaa100-1.16.128.160 16 131072 160 1 $2.71 Launch
rtx3090-4.16.128.160 16 131072 160 4 $3.23 Launch
rtx4090-4.16.128.160 16 131072 160 4 $4.26 Launch
rtx5090-3.16.96.160 16 98304 160 3 $4.34 Launch
teslah100-1.16.128.160 16 131072 160 1 $5.23 Launch

Related models

Qwen3-235B-A22B
Qwen3-0.6B
Qwen3-1.7B
Qwen3-4B
Qwen3-8B
Qwen3-14B
Qwen3-32B
DeepSeek-R1-0528-Qwen3-8B
T-pro-2.0
Qwen3-235B-A22B-Instruct-2507
Qwen3-235B-A22B-Thinking-2507
Qwen3-Coder-480B-A35B-Instruct
Qwen3-Coder-30B-A3B-Instruct

Need help?

Contact our dedicated neural networks support team at nn@immers.cloud or send your request to the sales department at sale@immers.cloud.