FitMyLLM
NVIDIA· Volta

NVIDIA Tesla V100 SXM3 32 GB

Running LLMs on the Tesla V100 SXM3 32 GB — the long read: which models fit at which quantisation, and the settings worth changing. · Or what a budget buys

VRAM
32 GB
High-end
Bandwidth
981
GB/s
Models Q4
362/498
73%
7B Q4 speed
~112
Blazing
Model coverage at Q473% of all
Estimated speed· By model size at Q4

Average speeds at Q4 quantization. Actual performance varies by model architecture and context length.

3B
~262
tok/s
7B
~112
tok/s
14B
~56
tok/s
32B
~25
tok/s
70B
—
39.4GB needed
Measured rig reports

We rent the machine and time every model on it: decode, VRAM peak, concurrency, watts and cost per million tokens. Yours may already be one of them — and two are free to read in full.

See the reports →
Specifications
VRAM
32 GB
Bandwidth
981 GB/s
FP16 Compute
32.7 TFLOPS
TDP
250W
Memory
HBM2
Architecture
Volta
CUDA Cores
5,120
Tensor Cores
640
PCIe
Gen 3 x16
313
Fast models · >30 tok/s
Real-time chat speed
362
Usable · >10 tok/s
Comfortable for all tasks
362
Total compatible
Fit in VRAM at Q4
Rent it in the cloud

Buying Tesla V100 SXM3 32 GB costs $15–$40k and isn’t practical for most teams. Spin one up by the hour instead:

Spin up in ~60s. Pay by the second. Cancel anytime.

Some links are affiliate links — we may earn a small commission at no extra cost to you. This helps keep FitMyLLM free and independent.

Compatible models· 362
S
Falcon-H1R Tiny 90M0.09B
Falcon·256K ctx· chat· reasoning
8720
tok/s · 2% VRAM
›
S
SmolLM2 135M0.135B
Smollm·2K ctx· chat
5813
tok/s · 2% VRAM
›
S
nomic-embed-text-v1.5 100M0.14B
Embedding·8K ctx· chat
5606
tok/s · 2% VRAM
›
S
GPT-2 124M0.14B
Gpt2·1K ctx· chat
5606
tok/s · 2% VRAM
›
S
SmolVLM 256M0.256B
Smollm·8K ctx· chat· vision
3066
tok/s · 2% VRAM
›
S
Gemma 3 270M0.27B
Gemma·32K ctx· chat
2907
tok/s · 2% VRAM
›
S
Snowflake Arctic Embed M v2.00.305B
Embedding·8K ctx· embedding· multilingual
2573
tok/s · 2% VRAM
›
S
bge-large-en-v1.5 335M0.335B
Embedding·1K ctx· chat
2343
tok/s · 2% VRAM
›
S
mxbai-embed-large-v10.335B
Embedding·1K ctx· embedding
2343
tok/s · 2% VRAM
›
S
Snowflake Arctic Embed L0.335B
Embedding·1K ctx· embedding
2343
tok/s · 2% VRAM
›
S
LFM2 350M0.35B
Lfm·125K ctx· chat· tool_use
2242
tok/s · 2% VRAM
›
S
SmolLM2 360M0.36B
Smollm·8K ctx· chat
2180
tok/s · 2% VRAM
›
S
GPT-2 Medium 345M0.38B
Gpt2·1K ctx· chat
2065
tok/s · 2% VRAM
›
S
Qwen 2.5 0.5B0.5B
Qwen·32K ctx· chat
1570
tok/s · 2% VRAM
›
S
SmolVLM 500M0.5B
Smollm·8K ctx· chat· vision
1570
tok/s · 2% VRAM
›
S
Falcon-H1 0.5B0.52B
Falcon·128K ctx· chat
1509
tok/s · 3% VRAM
›
S
BGE-M30.568B
Embedding·8K ctx· embedding
1382
tok/s · 3% VRAM
›
S
Snowflake Arctic Embed L v2.00.568B
Embedding·8K ctx· embedding· multilingual
1382
tok/s · 3% VRAM
›
S
Qwen3-Embedding 0.6B0.6B
Embedding·32K ctx· embedding· multilingual
1308
tok/s · 3% VRAM
›
S
Falcon-H1R Tiny 0.6B0.6B
Falcon·32K ctx· chat· reasoning
1308
tok/s · 3% VRAM
›
S
Falcon Perception 0.6B0.6B
Falcon·4K ctx· vision
1308
tok/s · 3% VRAM
›
S
Qwen 1.5 0.5B0.62B
Qwen·32K ctx· chat
1266
tok/s · 3% VRAM
›
S
LFM2 700M0.74B
Lfm·125K ctx· chat· tool_use
1061
tok/s · 3% VRAM
›
S
Qwen3 0.6B0.75B
Qwen·32K ctx· chat· reasoning
1046
tok/s · 3% VRAM
›
S
GPT-2 Large 774M0.81B
Gpt2·1K ctx· chat
969
tok/s · 3% VRAM
›
S
Qwen 3.5 0.8B0.87B
Qwen·256K ctx· chat· coding· multilingual
902
tok/s · 3% VRAM
›
S
InternVL3 1B0.94B
Other·32K ctx· chat· vision
835
tok/s · 3% VRAM
›
S
MiniCPM5 1B1.08B
Minicpm·128K ctx· chat· reasoning· multilingual
727
tok/s · 4% VRAM
›
S
K2 Horizon 0.9B1.08B
Other·128K ctx· chat· coding· reasoning
727
tok/s · 4% VRAM
›
S
TinyLlama 1.1B1.1B
Llama·2K ctx· chat
713
tok/s · 4% VRAM
›
S
LFM2.5-1.2B-Thinking1.2B
Lfm·122K ctx· chat· reasoning· tool_use
654
tok/s · 4% VRAM
›
S
Llama-3.2-1B1.2B
Llama·4K ctx· chat
654
tok/s · 4% VRAM
›
S
LFM2 1.2B1.2B
Lfm·125K ctx· chat· tool_use· multilingual
654
tok/s · 4% VRAM
›
S
Zamba2 1.2B1.2B
Other·4K ctx· chat
654
tok/s · 4% VRAM
›
S
Ling 3.0 Tiny7.89BMoE
Other·128K ctx· chat· coding· reasoning
604
tok/s · 17% VRAM
›
S
EXAONE-4.0-1.2B1.3B
Exaone·64K ctx· chat
604
tok/s · 4% VRAM
›
S
OPT 1.3B1.3B
Opt·2K ctx· chat
604
tok/s · 4% VRAM
›
S
MiniCPM-V 4.61.3B
Other·256K ctx· chat· vision
604
tok/s · 4% VRAM
›
S
DeepSeek Coder 1.3B1.35B
Deepseek·16K ctx· coding
581
tok/s · 4% VRAM
›
S
Phi-1 1.3B1.42B
Phi·2K ctx· coding
553
tok/s · 4% VRAM
›
Next step

Get personalized recommendations.

See ranked models with benchmark scores, run commands, and precise speed estimates for your Tesla V100 SXM3 32 GB.

What this card is worth

Tesla V100 SXM3 32 GB holds 362 of the models in our catalogue and is, in practice, a IQ4_XS card — the largest it takes is Nemotron-H 47B at IQ4_XS.

Tokens/sec per $100
—
8B at Q4_K_M, so cards compare like for like
VRAM per $100
—
what memory costs on this card
The biggest it takes
Nemotron-H 47B47B · IQ4_XS28.8 GB16 tok/sest
Phi-3.5 MoE 42B41.9B · Q4_K_M28.6 GB105 tok/sest
Falcon 40B40B · Q4_K_M27.4 GB17 tok/sest
Flint39.79B · Q4_K_M27.0 GB277 tok/sest
InternVL3 38B38B · Q4_K_M26.7 GB18 tok/sest
K2 Horizon MoVA 36B-A4B37.44B · Q4_K_M26.1 GB173 tok/sest
Seed-OSS 36B Instruct36B · Q5_K_S28.6 GB17 tok/sest
MiniCPM-V 4.7 35B-A3B35.21B · Q5_K_M27.7 GB198 tok/sest
Against cards of similar memory
M4 Max (48GB)36 GB$2,4992.2 tok/s per $100
M3 Max (48GB)36 GB$2,8991.6 tok/s per $100
M4 Pro (48GB)36 GB$1,7991.9 tok/s per $100
A100 SXM4 40 GB40 GB$10,0001.4 tok/s per $100

Neighbours in memory rather than in price: memory decides whether a card can do the job at all, so two cards of the same size at different prices is the comparison you are making. Ordered by memory, then bandwidth — not by the value column, which is worked out from bandwidth and price alone and therefore rewards a cheap card whatever its software stack does to that bandwidth in practice. Read it as one input, not as a ranking.

Device under test

NVIDIA Tesla V100 SXM3 32 GB — 32 GB VRAM.

Tesla V100 SXM3 32 GB spec
Brand
NVIDIA
VRAM
32 GB HBM2
Bandwidth
981 GB/s
FP16 Compute
32.7 TFLOPS
FP32 Compute
16.4 TFLOPS
CUDA Cores
5,120
Tensor Cores
640
TDP
250 W
Architecture
Volta
AI capability
362/ 498 models @ Q4

With 32 GB VRAM and 981 GB/s bandwidth, this GPU handles models up to 41.9B parameters.

Speed ≈ bandwidth / model_size × efficiency. A 7B model at Q4 runs at ~112 tok/s.

Rent it — live prices
Checking what the clouds are charging…
Top models for Tesla V100 SXM3 32 GB
362 fit · showing 20
ModelSizeVRAM Q4tok/sAvg
Phi-3.5 MoE 42B41.9B26.1 GB11956.7
Falcon 40B40B24.9 GB2020.9
Flint39.79B24.8 GB31477.3
InternVL3 38B38B23.7 GB2178.9
K2 Horizon MoVA 36B-A4B37.44B23.4 GB19645.7
Seed-OSS 36B Instruct36B22.5 GB2254.4
MiniCPM-V 4.7 35B-A3B35.21B22.0 GB26276.4
c4ai-command-r-v01 35B35B21.9 GB2227.5
Qwen 3.5 35B A3B35B21.9 GB26256.1
Qwen 3.6 35B A3B35B21.9 GB26253.3
K2 Horizon 32B34.78B21.7 GB2354.0
Nous Capybara 34B34.4B21.5 GB2342.0
Yi-1.5 34B34.4B21.5 GB2345.3
Falcon-H1 34B34B21.3 GB2358.6
CodeLlama 34B34B21.3 GB2325.4
Nous Hermes 2 34B34B21.3 GB2347.0
Phind CodeLlama 34B34B21.3 GB2368.1
LLaVA-1.6 Yi 34B34B21.3 GB2347.4
WizardCoder Python 34B34B21.3 GB2373.2
Yi 34B34B21.3 GB2333.4