Skip to content

Llama 4 Scout 17B 16E Instruct on 2× RTX 5090: VRAM, speed and fit

Llama 4 Scout 17B 16E Instruct at 3-bit runs on the 2× RTX 5090 with 48K of context at about 101 tokens a second.

Text generation · Chat · Ollama / LM Studio

Runs comfortably · on 2× RTX 5090 · 3-bit (Q3_K_M) · Ollama / LM Studio · Chat

GPU memory
60.2 GB
of 64.0 GB
Context
48K
auto
Speed
101 tok/s
57 when full
First reply
0.6 s
2K prompt

Where the memory goes

Model
49.5 GB
Context
9.0 GB
Overhead
1.7 GB
OS
0.0 GB
Free
3.8 GB
Memory
60.2 GB of 64.0 GB
Context
48K tokens · auto
Speed
101 tok/s · 57 when full
First reply
0.6 s for a 2K prompt
Running cost
$17/mo at home · $117/mo on a cloud 4× RTX 2000 Ada

Validation: Validated

What Llama 4 Scout 17B 16E Instruct on 2× RTX 5090 was checked against
CheckTargetResultStatus
Fits in GPU memory64.0 GB usable60.2 GB neededPass
Headroom5% free3.8 GB free (6%)Pass
Context for chat16K or more48KPass
Writing speed15 tok/s or more101 tok/sPass
First replyunder 3 s, 2K prompt0.6 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
Llama 4 Scout 17B 16E Instruct
Parameters
109B, 18B active
Precision
3-bit (Q3_K_M)
Context
48K, auto
Chats at once
1
Use
Chat
Runs in
Ollama / LM Studio
Hardware
2× RTX 5090
GPU can use
64.0 GB of 64.0 GB

Worth trying

  • 8-bit cache: Room for 96K context

Running cost

At home
$17/mo
770 W at the wall
Renting instead
$117/mo
4× RTX 2000 Ada at $0.96/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 unsloth/Llama-4-Scout-17B-16E-Instruct-GGUF
vLLM
vllm serve meta-llama/Llama-4-Scout-17B-16E-Instruct

Llama 4 Scout 17B 16E Instruct on similar machines

All machines: Llama 4 Scout 17B 16E Instruct VRAM requirements. Everything that runs here: what can the 2× RTX 5090 run?

Other models that run on 2× RTX 5090

Questions

Can Llama 4 Scout 17B 16E Instruct run on 2× RTX 5090?

Llama 4 Scout 17B 16E Instruct at 3-bit runs on the 2× RTX 5090 with 48K of context at about 101 tokens a second.

Report VR-0EP6IMC, calculated Nov 22, 2026 from published specs, not benchmarked.