Skip to content
4GB VRAM

RTX 3050 Ti Laptop GPU — what LLMs can it run?

27 of 151 indexed models fit comfortably in 4GB at 4K context, each at the highest-quality quant that still leaves headroom.

Context length
Concurrency (batch)

The short answer for RTX 3050 Ti Laptop GPU

4 GB, GDDR6, bandwidth not listed. 27 of 151 models in this index fit comfortably at 4K context; 29 load at all.

Hardware specifications source
Biggest that fits
Llama 3.1 8B Instruct — 3.3 GB at Q2_K — 0.7 GB spare at 4K, so longer context comes out of a thin margin.
Room to grow
Qwen2.5 3B Instruct — 2.1 GB at Q4_K_M — under 60% of the card, which leaves 1.9 GB for a long context window or a second process.
The wall you will hit
Laptop cooling and power limits affect performance even between laptops with the same GPU. Use the listed VRAM for fit, not desktop benchmark results for speed.

4–10B · 6

RTX 3050 Ti Laptop GPU — what LLMs can it run? — 4–10B
ModelQuantEst. VRAMHeadroom
Llama 3.1 8B Instruct8BQ2_K3.31 GB+0.7 GB
Gemma 4 E2B ITE2BQ4_K_M2.98 GB+1 GB
Qwen3.5 4B4BQ4_K_M3.29 GB+0.7 GB
Gemma 3 4B IT4BQ4_K_M3 GB+1 GB
Spark-X2.6 4B4BQ4_K_M3.01 GB+1 GB
Qwen3 4B Instruct4BEXL2 4.65bpw3.05 GB+1 GB

<4B · 21

RTX 3050 Ti Laptop GPU — what LLMs can it run? — <4B
ModelQuantEst. VRAMHeadroom
Phi-4 Mini Instruct3.8BQ4_K_M2.98 GB+1 GB
Granite 4.2 3B3BQ5_K_M3.07 GB+0.9 GB
Ministral 3 3B Instruct3BQ5_K_M3.03 GB+1 GB
Llama 3.2 3B Instruct3BQ4_K_M2.51 GB+1.5 GB
Qwen2.5 3B Instruct3BQ4_K_M2.11 GB+1.9 GB
SmolLM3 3B3BQ4_K_M2.26 GB+1.7 GB
Metis-1.6 Praxis3B MoEQ8_03.4 GB+0.6 GB
Gemma 2 2B Instruct2BQ8_03.34 GB+0.7 GB
Qwen3.5 2B2BQ8_02.23 GB+1.8 GB
LFM3 1.7B1.7BQ8_01.99 GB+2 GB
LFM3-VL 2B2BQ8_01.99 GB+2 GB
Qwen3 1.7B Instruct1.7BQ8_02.39 GB+1.6 GB
Qwen2.5 1.5B Instruct1.5BQ8_01.83 GB+2.2 GB
Llama 3.2 1B Instruct1BQ8_01.51 GB+2.5 GB
LFM2.5 1.2B Instruct1.2BQ8_01.35 GB+2.7 GB
K2 Horizon 0.9B0.9BQ8_01.44 GB+2.6 GB
Qwen3.5 0.8B0.8BQ8_00.94 GB+3.1 GB
BotPsy 0.8B0.75BQ8_00.9 GB+3.1 GB
Qwen2.5 0.5B Instruct0.5BQ8_00.6 GB+3.4 GB
MNN-Pocket 0.3B Instruct0.36BQ8_00.46 GB+3.5 GB
MNN-Pocket 0.3B QAT0.30BQ4_10.26 GB+3.7 GB

How this list is built

Each row is the lowest-perplexity-loss quant of that model whose estimated total — weights plus KV cache at 4K context plus activation buffer — uses at most 88% of the card. That is the calculator's "green" threshold, so each estimate leaves headroom. Raise the context length and the list shortens; the calculator lets you check any combination directly.

A further 2 models load but with no headroom to spare (up to 105% of VRAM) — the Quant Hub’s GPU chips count those too, which is why its number is higher.

Measured on this card

No inference runs were measured on this laptop GPU. Fit figures are estimates, not measured results.

Cards with the same budget

What fits is decided by memory, so every 4GB card of this type returns the same list. These pages are not different answers — they differ in throughput, which this index does not measure per card.

Stepping up

A RTX 4050 Laptop GPU (6GB) fits 19 more of the indexed models than this card. RTX 4050 Laptop GPU →

Common questions

What is the best local LLM for an RTX 3050 Ti Laptop GPU?

For everyday use, Qwen2.5 3B Instruct at Q4_K_M — about 2.1 GB of the card's 4 GB at 4K context, leaving 1.9 GB for a longer window. If you want the largest thing that will load, that is Llama 3.1 8B Instruct at Q2_K (3.3 GB, 0.7 GB spare). "Best" here means best fit for the memory budget — this index does not run task benchmarks, so it cannot tell you which model is smarter.

Can an RTX 3050 Ti Laptop GPU run Qwen3 8B Instruct?

No. Its smallest build here, AWQ INT4, needs about 4.9 GB and this card has 4.0 GB usable — short by 0.9 GB before any context beyond 4K. The largest model this card does clear is Llama 3.1 8B Instruct.

RTX 3050 Ti Laptop GPU or RTX 3050 Laptop GPU 4G for local LLMs?

They hold the same models: 4 GB against 4 GB fits 27 and 27 of 151 respectively at 4K. Bandwidth is not recorded for both cards here, so this page will not rank them on speed.

27 of 151 indexed models fit comfortably in 4GB at 4K context, each at the highest-quality quant that still leaves headroom.

This page's figures change when the model or the runtime does.Last updated 2026-11-22 RSS → /feed.xml