GPU VRAM Calculator

Which local LLM fits on your graphics card, card by card: usable VRAM, the largest model each card holds, and the best pick with its estimated speed.

What can my GPU run? Weights at Q4 cost about 0.6 GB per billion parameters, against a usable budget of roughly 90% of your VRAM. On the RTX 5080 (16 GB, the calculator's default tier), that is ~14 GB of weights, which the engine pairs with Gemma 4 E4B (Q8) at about ~95 tok/s (est.). Every card below is graded the same way: VRAM fixes the ceiling, bandwidth sets the speed, and the table shows what each one actually fits.

Requirements calculator
GB

With 16 GB of VRAM, ModelFit budgets about 14 GB for the model and comfortably runs local LLMs up to ~14B parameters at Q4. The best single pick is Gemma 4 E4B (Q8). Estimates use a representative 16 GB card (RTX 5080).

TOP PICKRuns wellA · 48

Gemma 4 E4B (Q8)

Gemma · top

Fits in 16 GB VRAM with room to spare. Best for on-device, mobile, chat on RTX 5080.

registry-verified
FOOTPRINT
7.5 GB
SPEED
~95 tok/s
FIRST TOKEN
~0.4s · Instant
QUANT
Q8_0
PARAMS
4.5B

Tokens/sec are ModelFit estimates from memory bandwidth and model size, not measured benchmarks. Ollama commands are registry-verified.

ALSO FITS
QWEN
Qwen3.5 9B Instruct (Q8)
Runs wellB · 26
SIZE
9B / Q8_0
LOAD
10.7 GB
SPEED
~53 tok/s
FIRST TOKEN
~0.4s · Instant
QWEN
Qwen3 8B (Q8)
Runs wellA · 44
SIZE
8B / Q8_0
LOAD
8.1 GB
SPEED
~58 tok/s
FIRST TOKEN
~0.4s · Instant
GEMMA
Gemma 4 12B
Runs wellA · 44
SIZE
12B / Q4_K_M
LOAD
8 GB
SPEED
~67 tok/s
FIRST TOKEN
~0.4s · Instant
QWEN
Qwen3 14B
Runs wellB · 24
SIZE
14B / Q4_K_M
LOAD
11 GB
SPEED
~58 tok/s
FIRST TOKEN
~0.4s · Instant
GEMMA
Gemma 3 12B Instruct
Runs wellA · 34
SIZE
12B / Q4_K_M
LOAD
9.5 GB
SPEED
~67 tok/s
FIRST TOKEN
~0.4s · Instant
LLAMA
Llama 3.1 8B Instruct (Q8)
Runs wellA · 44
SIZE
8B / Q8_0
LOAD
8 GB
SPEED
~58 tok/s
FIRST TOKEN
~0.4s · Instant
MISTRAL
Mistral Nemo 12B
Runs wellA · 34
SIZE
12B / Q4_K_M
LOAD
9.5 GB
SPEED
~67 tok/s
FIRST TOKEN
~0.4s · Instant

Every card, what it fits

One row per discrete card in ModelFit's catalog, computed by the same engine the calculator runs: the 90% weights budget, then the largest and best-scoring local models that fit. The GPU hub lists the same cards side by side.

CardVRAMUsable for weightsLargest model that fitsBest pick (est. tok/s)
RTX 40608 GB~7 GBOrnith 1.0 9B (9B)Gemma 4 E4B ~49 tok/s
RTX 306012 GB~11 GBGemma 4 12B (12B)Qwen3.5 9B Instruct ~38 tok/s
RTX 4060 Ti16 GB~14 GBLFM2 24B-A2B Instruct (24B, 2B active)Gemma 4 E4B (Q8) ~34 tok/s
RTX 5060 Ti16 GB~14 GBLFM2 24B-A2B Instruct (24B, 2B active)Gemma 4 E4B (Q8) ~52 tok/s
RTX 407012 GB~11 GBGemma 4 12B (12B)Qwen3.5 9B Instruct ~47 tok/s
RTX 4070 SUPER12 GB~11 GBGemma 4 12B (12B)Qwen3.5 9B Instruct ~51 tok/s
RTX 507012 GB~11 GBGemma 4 12B (12B)Qwen3.5 9B Instruct ~53 tok/s
RTX 5070 Ti16 GB~14 GBLFM2 24B-A2B Instruct (24B, 2B active)Gemma 4 E4B (Q8) ~88 tok/s
RTX 4070 Ti SUPER16 GB~14 GBLFM2 24B-A2B Instruct (24B, 2B active)Gemma 4 E4B (Q8) ~73 tok/s
RTX 4080 SUPER16 GB~14 GBLFM2 24B-A2B Instruct (24B, 2B active)Gemma 4 E4B (Q8) ~80 tok/s
RTX 508016 GB~14 GBLFM2 24B-A2B Instruct (24B, 2B active)Gemma 4 E4B (Q8) ~95 tok/s
RTX 309024 GB~22 GBQwen3.5 35B-A3B Instruct (35B, 3B active)Gemma 4 26B-A4B ~71 tok/s
RTX 409024 GB~22 GBQwen3.5 35B-A3B Instruct (35B, 3B active)Gemma 4 26B-A4B ~85 tok/s
RTX 509032 GB~29 GBQwen3.6 35B-A3B (35B, 3B active)Qwen3.6 35B-A3B ~118 tok/s
RTX PRO 600096 GB~86 GBQwen3.5 122B-A10B Instruct (122B, 10B active)GPT-OSS 120B ~51 tok/s
RX 7900 XTX24 GB~22 GBQwen3.5 35B-A3B Instruct (35B, 3B active)Gemma 4 26B-A4B ~72 tok/s
RX 7900 XT20 GB~18 GBMuse Glimmer 30B (29.8B)GPT-OSS 20B ~62 tok/s

Q4_K_M builds from ModelFit's 106-model catalog. Fit and tok/s are ModelFit estimates from memory bandwidth and model size, not measured benchmarks. Updated 2026-09-18.

Speed by model size on three popular cards

These are bandwidth-derived estimates, not measured benchmarks: each card's rate scales from its own memory bandwidth and the 90% usable budget. Rows marked CPU offload exceed the card's VRAM; MoE rows scale by active parameters, so they degrade far less when offloaded.

RTX 4060 8 GB · 272 GB/s
Model sizeEst. tok/sFits?
7B~34 tok/sFits in VRAM
14B~4 tok/sCPU offload (slow)
20B MoE (3.6B active) (MoE)~10 tok/sCPU offload (slow)
32B~1 tok/sCPU offload (slow)
35B MoE (3B active) (MoE)~9 tok/sCPU offload (slow)
70Bnot practicalCPU offload (slow)
120B MoE (5.1B active) (MoE)~4 tok/sCPU offload (slow)
RTX 4090 24 GB · 1008 GB/s
Model sizeEst. tok/sFits?
7B~117 tok/sFits in VRAM
14B~65 tok/sFits in VRAM
20B MoE (3.6B active) (MoE)~99 tok/sFits in VRAM
32B~32 tok/sFits in VRAM
35B MoE (3B active) (MoE)~84 tok/sFits in VRAM
70B~2 tok/sCPU offload (slow)
120B MoE (5.1B active) (MoE)~14 tok/sCPU offload (slow)
RTX 5090 32 GB · 1792 GB/s
Model sizeEst. tok/sFits?
7B~162 tok/sFits in VRAM
14B~90 tok/sFits in VRAM
20B MoE (3.6B active) (MoE)~138 tok/sFits in VRAM
32B~45 tok/sFits in VRAM
35B MoE (3B active) (MoE)~117 tok/sFits in VRAM
70B~23 tok/sFits in VRAM
120B MoE (5.1B active) (MoE)~19 tok/sCPU offload (slow)

Anchored to an 8B-class Q4_K_M model and scaled by model size; estimates, not measured benchmarks. Updated 2026-09-18.

How to read your card's spec sheet

A graphics card spec sheet lists dozens of numbers, but two decide everything about local LLMs. VRAM is the hard ceiling: at Q4 the weights cost about 0.6 GB per billion parameters, and ModelFit budgets 90% of the card for them, which is why the table above rounds every card to its real usable budget. Memory bandwidth sets the speed once the model fits, because wider memory moves tokens out faster, and every tok/s figure on this page is an estimate scaled from it. Clock speeds and core counts move performance by a few percent; these two numbers decide whether the model runs at all and how it feels while it does.

Frequently asked questions

Is 8GB of VRAM enough for local LLMs in 2026?

Eight gigabytes still covers the small-model tier. The 90% usable budget leaves about 7 GB for weights, which holds an 8B model at Q4 (5 GB) with modest context and powers lightweight chat and coding assistants; the RTX 4060 row of the table above shows exactly what the engine pairs with it. A 14B model needs about 8 GB of weights, so an 8GB card only reaches it with heavier quantization or CPU offload, both of which cost quality and speed. If 14B-class is your floor, a 12GB or 16GB card is the more comfortable starting point.

Does memory bandwidth matter as much as VRAM?

They answer different questions. VRAM decides what fits: nothing runs if the weights do not fit the budget. Bandwidth decides how fast it runs once it does, because decode speed is roughly proportional to it for a given model size, which is why every tok/s figure on this page is bandwidth-derived. The practical consequence is that a faster older card with less VRAM can beat a roomier one on small models, so the best-pick column weighs fit first and speed second. Mixture-of-experts models bend the rule in your favor: only their active experts are read per token, so decode speed tracks active parameters, not the total.

Which GPU do I need for a 32B model?

A dense 32B at Q4 needs about 19 GB of weights, and against the 90% budget that implies a card with at least 22 GB of VRAM. In practice that means a 24GB card, such as the RTX 3090, RTX 4090 or RX 7900 XTX, which all hold it comfortably per the table above. A 16GB card lands at about 14 GB usable, roughly 5 GB short of a dense 32B, so it falls back to lower quantizations or CPU offload.

Can I pool two GPUs to run a bigger model?

Yes, and that is exactly what the multi-GPU rig mode of the calculator below computes. Two 16GB cards pool about 29 GB of usable weight budget, which clears the 19 GB a dense 32B needs. The honest caveat is that pooled throughput is lower than a single card with the same total VRAM, because layers are transferred between cards, and mixture-of-experts models tolerate pooling better since only their active experts cross the link. Pooling is the practical path to 70B-class models on consumer cards, which is why the rig mode exists.

Go deeper

More surfaces: all GPU pages · can I run it · best hardware for a model · quantization comparison · open compatibility dataset · hardware stats

The weekly local-AI refresh

New open-weight models, real Apple Silicon benchmarks, and the one model worth running on your Mac this week. Free, one email a week, unsubscribe anytime.

By subscribing you agree to our Privacy Policy and to receive the weekly email. Unsubscribe anytime.