Is 8GB of VRAM enough for AI?

It depends entirely on which models, and the honest answer is a list rather than a yes or no. On our measured gates, 8GB runs 3 of our 12 workloads: Qwen3 4B, Llama 3.1 8B, Stable Diffusion XL. It is gated out of 9: Qwen2.5-Coder 14B, Z-Image Turbo, LTX-Video, Wan 2.2 (720p), Qwen3 32B, FLUX.1-dev, FLUX.1 Kontext, Llama 3.3 70B, Qwen-Image-Edit. The nearest miss is Qwen2.5-Coder 14B, which needs about 11.5GB, short by 3.5GB.

Can you run Llama 3.3 70B on 8GB?

No. Llama 3.3 70B needs roughly 42GB at Q4_K_M and a 8GB card is 34GB short. This isn't a speed problem, the weights don't fit, so the model doesn't load. The largest LLM in our ladder that 8GB does run is Llama 3.1 8B.

Can you run FLUX.1-dev on 8GB?

No. FLUX.1-dev needs roughly 26GB at BF16 and a 8GB card is 18GB short. This is the cliff that catches most people out, because 8GB handles SDXL without complaint and feels like plenty right up until you try to load a modern image model. Running FLUX at a smaller quantisation is possible but it isn't the same model, which is why we don't publish it in the same column.

What's the next VRAM tier worth stepping up to from 8GB?

11.5GB, and what it buys you is specifically Qwen2.5-Coder 14B. That's the useful way to think about a VRAM upgrade: not as a percentage more headroom, but as a named list of models you couldn't run before and can now. Stepping up a tier at the same bandwidth gets you capability. Buying a faster card at the same capacity gets you a percentage.

Does a smaller quantisation get me past the VRAM limit?

Often, yes, and we deliberately don't count it. Running a model at a lower quantisation reduces its memory requirement and its output quality, and if we silently swapped precision whenever a card ran out of room, every number on this site would become meaningless. Our floors are for Q4_K_M language models and BF16 diffusion. If you drop below that you're running a different model, and it should be measured as one.

Do I need system RAM as well as 8GB of VRAM?

For some workloads, yes, and it catches people out. LTX-Video wants roughly 20GB of host RAM and Wan 2.2 wants around 38GB, on top of their VRAM requirement, because the weights have to be staged. A card with enough VRAM sitting in a RAM-starved machine still fails. We gate on both, which is why a few cards you'd expect to clear a workload don't.

I'm between two VRAM tiers and money's tight. Which way do I jump?

Here's my rule after benchmarking the whole board: between 16GB and 24GB, go 16 and rent for the big jobs, the jump to 24 isn't big enough to matter once you account for context-window headroom on top of model weights, so they're effectively the same tier now. The real step up is 32GB on the newest architecture (RTX 5090). Below that, buy the cheaper card and put the savings into rented hours on cards you could never justify owning.

VRAM tiers · measured gates across our fleet · Updated July 2026

What Can You Actually Run on 8GB of VRAM?

A 8GB GPU runs 3 of the 12 AI workloads we benchmark and is locked out of 9. 8GB is the entry floor for local AI, and it is a real floor rather than a soft one. You get small language models and classic image generation, and that is genuinely most of what people do, but every modern diffusion model and everything above a 14B parameter LLM is simply out of reach. Every floor below is measured: we ran each model on each card until it failed, and published where it failed.

3/12

Workloads that run

at the precision we test

9/12

Workloads gated out

hard VRAM ceiling, no fix

11.5GB

Next thing you'd unlock

Qwen2.5-Coder 14B

8GB cards in our data

of the fleet we've benchmarked

8GB is the entry floor for local AI, and it is a real floor rather than a soft one. You get small language models and classic image generation, and that is genuinely most of what people do, but every modern diffusion model and everything above a 14B parameter LLM is simply out of reach.

What runs on 8GB. Measured floors from our own gate testing

Model	Type	VRAM floor	On 8GB?
Qwen3 4B	LLM	~5GB	Runs
Llama 3.1 8B	LLM	~8GB	Runs
Stable Diffusion XL	Image	~8GB	Runs
Qwen2.5-Coder 14B	LLM	~11.5GB	Won't fit (short 3.5GB)
Z-Image Turbo	Image	~13GB	Won't fit (short 5GB)
LTX-Video	Video	~14GB	Won't fit (short 6GB)
Wan 2.2 (720p)	Video	~18GB	Won't fit (short 10GB)
Qwen3 32B	LLM	~20GB	Won't fit (short 12GB)
FLUX.1-dev	Image	~26GB	Won't fit (short 18GB)
FLUX.1 Kontext	Editing	~26GB	Won't fit (short 18GB)
Llama 3.3 70B	LLM	~42GB	Won't fit (short 34GB)
Qwen-Image-Edit	Editing	~42GB	Won't fit (short 34GB)

Floors are measured at Q4_K_M for language models and BF16 for diffusion (SDXL at FP16). A smaller quantisation lowers the requirement and the quality, we don't mix precisions in one column.

Two things decide whether a GPU can do AI, and people consistently get the order wrong. Capacity decides what runs. Bandwidth decides how fast it runs. Capacity comes first, because a model that doesn't fit doesn't run slowly: it doesn't run at all, and no amount of bandwidth rescues it. That's why our AI Score treats a won't-fit as a zero rather than quietly excluding it. The practical consequence: shop for the VRAM tier that clears the models you actually intend to use, then optimise for bandwidth within that tier. Buying a faster card at the same capacity gets you a percentage. Buying the next capacity tier up can get you a model you literally could not run before.

Our verdict

8GB runs a useful assistant and SDXL. It does not run modern image models, and no setting changes that. On our measured gates, 8GB clears 3 of 12 workloads and misses 9, the nearest being Qwen2.5-Coder 14B at ~11.5GB, short by 3.5GB. Buy capacity first and bandwidth second: capacity decides what runs, bandwidth only decides how fast.

FAQ

Is 8GB of VRAM enough for AI?: It depends entirely on which models, and the honest answer is a list rather than a yes or no. On our measured gates, 8GB runs 3 of our 12 workloads: Qwen3 4B, Llama 3.1 8B, Stable Diffusion XL. It is gated out of 9: Qwen2.5-Coder 14B, Z-Image Turbo, LTX-Video, Wan 2.2 (720p), Qwen3 32B, FLUX.1-dev, FLUX.1 Kontext, Llama 3.3 70B, Qwen-Image-Edit. The nearest miss is Qwen2.5-Coder 14B, which needs about 11.5GB, short by 3.5GB.
Can you run Llama 3.3 70B on 8GB?: No. Llama 3.3 70B needs roughly 42GB at Q4_K_M and a 8GB card is 34GB short. This isn't a speed problem, the weights don't fit, so the model doesn't load. The largest LLM in our ladder that 8GB does run is Llama 3.1 8B.
Can you run FLUX.1-dev on 8GB?: No. FLUX.1-dev needs roughly 26GB at BF16 and a 8GB card is 18GB short. This is the cliff that catches most people out, because 8GB handles SDXL without complaint and feels like plenty right up until you try to load a modern image model. Running FLUX at a smaller quantisation is possible but it isn't the same model, which is why we don't publish it in the same column.
What's the next VRAM tier worth stepping up to from 8GB?: 11.5GB, and what it buys you is specifically Qwen2.5-Coder 14B. That's the useful way to think about a VRAM upgrade: not as a percentage more headroom, but as a named list of models you couldn't run before and can now. Stepping up a tier at the same bandwidth gets you capability. Buying a faster card at the same capacity gets you a percentage.
Does a smaller quantisation get me past the VRAM limit?: Often, yes, and we deliberately don't count it. Running a model at a lower quantisation reduces its memory requirement and its output quality, and if we silently swapped precision whenever a card ran out of room, every number on this site would become meaningless. Our floors are for Q4_K_M language models and BF16 diffusion. If you drop below that you're running a different model, and it should be measured as one.
Do I need system RAM as well as 8GB of VRAM?: For some workloads, yes, and it catches people out. LTX-Video wants roughly 20GB of host RAM and Wan 2.2 wants around 38GB, on top of their VRAM requirement, because the weights have to be staged. A card with enough VRAM sitting in a RAM-starved machine still fails. We gate on both, which is why a few cards you'd expect to clear a workload don't.
I'm between two VRAM tiers and money's tight. Which way do I jump?: Here's my rule after benchmarking the whole board: between 16GB and 24GB, go 16 and rent for the big jobs, the jump to 24 isn't big enough to matter once you account for context-window headroom on top of model weights, so they're effectively the same tier now. The real step up is 32GB on the newest architecture (RTX 5090). Below that, buy the cheaper card and put the savings into rented hours on cards you could never justify owning.

How we test

The VRAM floors on this page are measured, not calculated. We ran every model in our suite on every GPU in our fleet and recorded where it failed. When a model exceeds a card's VRAM we publish a hard won't-fit result with the requirement we observed, rather than quietly dropping to a smaller quantisation to produce a number, a card that can't run a model scores zero on it in our AI Score. LLMs are measured at Q4_K_M on llama.cpp (llama-bench, -p 512 -n 128). Diffusion and video run at BF16 on diffusers/ComfyUI, with SDXL at FP16. Floors are for those precisions: running a model at a smaller quantisation will lower its VRAM requirement and its quality, and we don't mix the two in one column. Two caveats worth stating plainly. First, these floors assume the model is the only thing on the card: a display attached to the same GPU, or a desktop compositor, eats into your headroom. Second, some workloads gate on system RAM as well as VRAM, LTX-Video wants roughly 20GB of host RAM and Wan 2.2 wants around 38GB, and a card with enough VRAM in a RAM-starved machine still fails. We gate on both. Speed figures are single-GPU, single-stream, batch-size-1, from our own runs. Cards labelled Measured were rented and run by us; Estimated cards are interpolated per workload against those anchors and labelled on every row.