Can LLaVA 1.6 13B run on RTX 5090 32GB?

YES — Runs Great

A80Great

Estimated from fit model

LLaVA 1.6 13B needs ~24.5 GB VRAM. RTX 5090 32GB has 32.0 GB. With Q4_K_M quantization, expect ~151 tok/s.

Runtime: OllamaCapacity: RoomyBandwidth: HighStack: BasicBottleneck: Balanced

Operating mode

Choose the run profile you care about

Interactive favors responsiveness, while light API and scale-out lean harder on serving readiness. The fit stays the same, but the recommendation lens changes.

Current mode

Balanced

Balanced for general local use. Keeps the ranking neutral across personal and serving workflows.

Capabilities:

Select quantization to explore

Q4_K_M (Medium quality) — 24.5 GB, 151.4 tok/s, Runs well

24.5 GB required32.0 GB available

77% VRAM used

Fit status

Runs well

Decode

151.4 tok/s

TTFT

1279 ms

Safe context

Memory

24.5 GB / 32.0 GB

Memory breakdown

Weights7.9 GB

KV Cache12.2 GB

Runtime1.2 GB

Headroom3.2 GB

See how fast it feels

See how fast it feelsLLaVA 1.6 13B on RTX 5090 32GB
1st promptCold start — includes initialization
>What is local AI inference?
Local AI inference means running an AI language model directly on your own hardware — your laptop, desktop, or server — instead of sending requests to a remote cloud API.

When you run inference locally the model weights are loaded into your GPU or unified memory. Each token you generate requires reading those weights from memory, so memory bandwidth is the main bottleneck for decode speed.

Key benefits of running locally:
  - Full privacy: your prompts never leave your machine
  - No per-token cost or rate limits
  - Works offline once the model is downloaded
  - Latency depends only on your hardware
2nd promptWarm — model ready, no init overhead
>How much VRAM do I need?
It depends on the model size and quantization level. A rough rule of thumb:

  Model size     Q4 (4-bit)    Q8 (8-bit)    FP16
  7B params      ~4.3 GB       ~7.5 GB       ~14 GB
  13B params     ~7.9 GB       ~13.9 GB      ~26 GB
  70B params     ~42.7 GB      ~74.9 GB      ~140 GB

Most people use 4-bit quantization (Q4_K_M) which gives 90-95% of full quality at a fraction of the memory. A 24 GB GPU can comfortably run most 7B-13B models.
Estimated: 151.4 tok/s decode · 1.3s TTFT (warm) · 379 tok/s prefill

What limits this setup

This setup is broadly balanced for this model.

No major red flags

This recommendation has enough memory headroom and acceptable estimated speed for the selected workload.

Best improvement path

Performance by workload

Workload	Grade	Fit	Decode	TTFT	Context
Chat	A	Runs well	151.4 tok/s	697 ms	4K
Coding	A	Runs well	151.4 tok/s	1279 ms	4K
Agentic Coding	B	Very compromised (needs ~1 GB host RAM)	87.3 tok/s	3225 ms	4K
Reasoning	A	Runs well	151.4 tok/s	1511 ms	4K
RAG	B	Very compromised (needs ~1 GB host RAM)	87.3 tok/s	4031 ms	4K

Inference speed

LLaVA 1.6 13B inference speed — tokens per second by GPU & Mac

Estimated decode speed (tokens/sec) for LLaVA 1.6 13B at Q4_K_M across popular GPUs and Apple Silicon, using the fastest local runtime per device. Fastest is RTX 5090 32GB at ~151 tok/s. Speed is memory-bandwidth bound, so cards that fit the whole model in VRAM run far faster than ones that offload to system RAM.

GPU / Mac	Memory	Quant	Speed (tok/s)	Fits?
RTX 5090 32GB	32 GB	Q4_K_M	151.4	Fits
RTX 4090 24GB	24 GB	Q4_K_M	96.6	Offloads
RX 7900 XTX 24GB	24 GB	Q4_K_M	87.2	Offloads
RTX 3090 24GB	24 GB	Q4_K_M	82.6	Offloads
Mac Studio M3 Ultra 256GB	256 GB	Q4_K_M	70.2	Fits
Mac Studio M2 Ultra 128GB	128 GB	Q4_K_M	58.5	Fits
Mac Studio M1 Ultra 128GB	128 GB	Q4_K_M	55.5	Fits
MacBook Pro M4 Max 128GB	128 GB	Q4_K_M	38.2	Fits
MacBook Pro M4 Max 64GB	64 GB	Q4_K_M	38.2	Fits
MacBook Pro M3 Max 64GB	64 GB	Q4_K_M	30.3	Fits
MacBook Pro M1 Max 64GB	64 GB	Q4_K_M	27.7	Fits
RTX 4080 Super 16GB	16 GB	Q4_K_M	27.1	Too big
MacBook Pro M4 Pro 48GB	48 GB	Q4_K_M	23.3	Fits
RTX 4070 12GB	12 GB	Q4_K_M	9.5	Too big
RTX 3060 12GB	12 GB	Q4_K_M	6.0	Too big
RTX 4060 8GB	8 GB	Q4_K_M	3.8	Too big

Estimates for single-stream decoding at Q4_K_M; real tokens/sec varies with prompt length, context, batch size, and runtime build. Prompt processing (prefill) is faster than the decode figures shown here.

Quantization options

How LLaVA 1.6 13B (13B params) fits at each quantization level on RTX 5090 32GB (32.0 GB usable).

Quant	Bits	VRAM	Quality	Fit
Q2_K	2	5.1 GB	Low	B67
Q3_K_S	3	6.4 GB	Low	B68
NVFP4	4	7.3 GB	Medium	B68
Q4_K_M	4	7.9 GB	Medium	B68
Q5_K_M	5	9.4 GB	High	B69
Q6_K	6	10.7 GB	High	B69
Q8_0	8	13.9 GB	Very High	A71
F16Best for your GPU	16	26.7 GB	Maximum	A72

Get started

Copy-paste commands to run LLaVA 1.6 13B on your machine.

Run

docker run --rm -it ghcr.io/ggerganov/llama.cpp:full \
  --hf-repo "liuhaotian/llava-v1.6-mistral-7b" \
  --hf-file "llava-v1.6-mistral-7b-Q4_K_M.gguf" \
  -c 4096 -ngl 99

Your hardware

More models your RTX 5090 32GB can run

Model	Params	Grade	Decode
Qwen3-Coder 30B A3B Instruct	30.5B	S	181.6 tok/s
Qwen 3.5 27B	27B	S	78.7 tok/s
Qwen 3.6 27B	27B	S	79 tok/s
Qwen 3.6 35B A3B	35B	S	128.2 tok/s
Qwen3-VL 30B A3B Instruct	30B	S	187.8 tok/s

Frequently asked questions

Can RTX 5090 32GB run LLaVA 1.6 13B?

Yes, RTX 5090 32GB can run LLaVA 1.6 13B with a A grade (Runs well). Expected decode speed: 151.4 tok/s.

How much VRAM does LLaVA 1.6 13B need?

LLaVA 1.6 13B (13B parameters) requires approximately 24.5 GB of memory with Q4_K_M quantization.

What is the best quantization for LLaVA 1.6 13B?

The recommended quantization for LLaVA 1.6 13B is Q4_K_M, which balances quality and memory efficiency.

What speed will LLaVA 1.6 13B run at on RTX 5090 32GB?

On RTX 5090 32GB, LLaVA 1.6 13B achieves approximately 151.4 tokens per second decode speed with a time-to-first-token of 1279ms using Q4_K_M quantization.

Can RTX 5090 32GB run LLaVA 1.6 13B for coding?

For coding workloads, LLaVA 1.6 13B on RTX 5090 32GB receives a A grade with 151.4 tok/s and 4K context.

What context window can LLaVA 1.6 13B use on RTX 5090 32GB?

On RTX 5090 32GB, LLaVA 1.6 13B can safely use up to 4K tokens of context. The model's official context limit is 4K, but available memory constrains the safe maximum.

See all results for RTX 5090 32GB See all hardware for LLaVA 1.6 13B

Embed this result▼

Paste this snippet into any page to show a live fit card.

<iframe src="https://willitrunai.com/embed/llava-1.6-13b-on-rtx-5090-32gb" width="400" height="180" frameborder="0" style="border:none;border-radius:12px;overflow:hidden;" title="Will It Run AI — fit result"></iframe>

Preview:

Workload

Grade

Fit

Decode

TTFT

Context

Chat

Runs well

151.4 tok/s

697 ms

Coding

Runs well

151.4 tok/s

1279 ms

Agentic Coding

Very compromised (needs ~1 GB host RAM)

87.3 tok/s

3225 ms

Reasoning

Runs well

151.4 tok/s

1511 ms

RAG

Very compromised (needs ~1 GB host RAM)

87.3 tok/s

4031 ms

LLaVA 1.6 13B inference speed — tokens per second by GPU & Mac

GPU / Mac

Memory

Quant

Speed (tok/s)

Fits?

RTX 5090 32GB

32 GB

Q4_K_M

151.4

Fits

RTX 4090 24GB

24 GB

Q4_K_M

96.6

Offloads

RX 7900 XTX 24GB

24 GB

Q4_K_M

87.2

Offloads

RTX 3090 24GB

24 GB

Q4_K_M

82.6

Offloads

Mac Studio M3 Ultra 256GB

256 GB

Q4_K_M

70.2

Fits

Mac Studio M2 Ultra 128GB

128 GB

Q4_K_M

58.5

Fits

Mac Studio M1 Ultra 128GB

128 GB

Q4_K_M

55.5

Fits

MacBook Pro M4 Max 128GB

128 GB

Q4_K_M

38.2

Fits

MacBook Pro M4 Max 64GB

64 GB

Q4_K_M

38.2

Fits

MacBook Pro M3 Max 64GB

64 GB

Q4_K_M

30.3

Fits

MacBook Pro M1 Max 64GB

64 GB

Q4_K_M

27.7

Fits

RTX 4080 Super 16GB

16 GB

Q4_K_M

27.1

Too big

MacBook Pro M4 Pro 48GB

48 GB

Q4_K_M

23.3

Fits

RTX 4070 12GB

12 GB

Q4_K_M

9.5

Too big

RTX 3060 12GB

12 GB

Q4_K_M

6.0

Too big

RTX 4060 8GB

8 GB

Q4_K_M

3.8

Too big

Quantization options

How LLaVA 1.6 13B (13B params) fits at each quantization level on RTX 5090 32GB (32.0 GB usable).

Quant	Bits	VRAM	Quality	Fit
Q2_K	2	5.1 GB	Low	B67
Q3_K_S	3	6.4 GB	Low	B68
NVFP4	4	7.3 GB	Medium	B68
Q4_K_M	4	7.9 GB	Medium	B68
Q5_K_M	5	9.4 GB	High	B69
Q6_K	6	10.7 GB	High	B69
Q8_0	8	13.9 GB	Very High	A71
F16Best for your GPU	16	26.7 GB	Maximum	A72

Model

Params

Grade

Decode

Capabilities

Qwen3-Coder 30B A3B Instruct

30.5B

181.6 tok/s

Qwen 3.5 27B

27B

78.7 tok/s

Qwen 3.6 27B

27B

79 tok/s

Qwen 3.6 35B A3B

35B

128.2 tok/s

Qwen3-VL 30B A3B Instruct

30B

187.8 tok/s

Frequently asked questions

Can RTX 5090 32GB run LLaVA 1.6 13B?

Yes, RTX 5090 32GB can run LLaVA 1.6 13B with a A grade (Runs well). Expected decode speed: 151.4 tok/s.

How much VRAM does LLaVA 1.6 13B need?

LLaVA 1.6 13B (13B parameters) requires approximately 24.5 GB of memory with Q4_K_M quantization.

What is the best quantization for LLaVA 1.6 13B?

The recommended quantization for LLaVA 1.6 13B is Q4_K_M, which balances quality and memory efficiency.

What speed will LLaVA 1.6 13B run at on RTX 5090 32GB?

On RTX 5090 32GB, LLaVA 1.6 13B achieves approximately 151.4 tokens per second decode speed with a time-to-first-token of 1279ms using Q4_K_M quantization.

Can RTX 5090 32GB run LLaVA 1.6 13B for coding?

For coding workloads, LLaVA 1.6 13B on RTX 5090 32GB receives a A grade with 151.4 tok/s and 4K context.

What context window can LLaVA 1.6 13B use on RTX 5090 32GB?

On RTX 5090 32GB, LLaVA 1.6 13B can safely use up to 4K tokens of context. The model's official context limit is 4K, but available memory constrains the safe maximum.