Skip to content

Qwen3 4B on RTX 3090: VRAM, speed and fit

Qwen3 4B at 8-bit runs on the RTX 3090 with 40K of context at about 132 tokens a second.

Text generation · Chat · Ollama / LM Studio

Runs comfortably · on RTX 3090 · 8-bit (Q8_0) · Ollama / LM Studio · Chat

GPU memory
10.4 GB
of 24.0 GB
Context
40K
auto
Speed
132 tok/s
59 when full
First reply
0.2 s
2K prompt

Where the memory goes

Model
4.0 GB
Context
5.6 GB
Overhead
838 MB
OS
0.0 GB
Free
13.6 GB
Memory
10.4 GB of 24.0 GB
Context
40K tokens · auto
Speed
132 tok/s · 59 when full
First reply
0.2 s for a 2K prompt
Running cost
$6.35/mo at home · $29/mo on a cloud RTX 2000 Ada

Validation: Validated

What Qwen3 4B on RTX 3090 was checked against
CheckTargetResultStatus
Fits in GPU memory24.0 GB usable10.4 GB neededPass
Headroom5% free13.6 GB free (57%)Pass
Context for chat16K or more40KPass
Writing speed15 tok/s or more132 tok/sPass
First replyunder 3 s, 2K prompt0.2 sPass

Targets: a reply you can read along with, a first reply the use case tolerates, and 5% of memory left free. Calculated from published specs, not benchmarked.

The setup

Model
Qwen3 4B
Parameters
4B
Precision
8-bit (Q8_0)
Context
40K, auto
Chats at once
1
Use
Chat
Runs in
Ollama / LM Studio
Hardware
RTX 3090
GPU can use
24.0 GB of 24.0 GB

Running cost

At home
$6.35/mo
290 W at the wall
Renting instead
$29/mo
RTX 2000 Ada at $0.24/h

4 h a day, power at $0.18 per kWh, cloud at RunPod on-demand prices as of Nov 22, 2026.

How to run it

llama.cpp
llama-server -hf Qwen/Qwen3-4B-GGUF
vLLM
vllm serve Qwen/Qwen3-4B
MLX (Mac)
mlx_lm.chat --model mlx-community/Qwen3-4B-4bit

Qwen3 4B on similar machines

  • RTX 4080 SuperRuns
  • RTX 5070 TiRuns
  • RTX 5080Runs
  • Radeon RX 7900 XTXRuns
  • RTX 4090Runs
  • Radeon AI PRO R9700Runs
  • RTX 5090Runs
  • 2× RTX 3090Runs

All machines: Qwen3 4B VRAM requirements. Everything that runs here: what can the RTX 3090 run?

Other models that run on RTX 3090

Questions

Can Qwen3 4B run on RTX 3090?

Qwen3 4B at 8-bit runs on the RTX 3090 with 40K of context at about 132 tokens a second.

Report VR-1BZZQ1C, calculated Nov 22, 2026 from published specs, not benchmarked.