FitMyLLM
NVIDIA· KEPLER

NVIDIA NVS 510

Running LLMs on the NVS 510 — the long read: which models fit at which quantisation, and the settings worth changing. · Or what a budget buys

VRAM
2 GB
BUDGET
BANDWIDTH
29
GB/S
MODELS Q4
47/449
10%
7B Q4 SPEED
~3
SLOW
▸ MODEL COVERAGE @ Q410% OF ALL
▸ ESTIMATED SPEED· BY MODEL SIZE @ Q4

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

3B
1.7GB NEEDED
7B
3.9GB NEEDED
14B
7.9GB NEEDED
32B
18.0GB NEEDED
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
2 GB
BANDWIDTH
29 GB/s
FP16 COMPUTE
0.3 TFLOPS
TDP
35W
MEMORY
DDR3
ARCHITECTURE
Kepler
CUDA CORES
192
PCIE
Gen 2 x16
24
FAST MODELS · >30 TOK/S
Real-time chat speed
47
USABLE · >10 TOK/S
Comfortable for all tasks
47
TOTAL COMPATIBLE
Fit in VRAM at Q4
▸ DON’T WANT TO BUY?

Test NVS 510 (or anything bigger) without committing. Pay by the second, cancel anytime.

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· 47
S
Falcon-H1R Tiny 90M0.09B
FALCON·256K CTX· CHAT· REASONING
258
TOK/S · 27% VRAM
S
SmolLM2 135M0.135B
SMOLLM·2K CTX· CHAT
172
TOK/S · 29% VRAM
S
nomic-embed-text-v1.5 100M0.14B
EMBEDDING·8K CTX· CHAT
166
TOK/S · 29% VRAM
S
GPT-2 124M0.14B
GPT2·1K CTX· CHAT
166
TOK/S · 29% VRAM
S
SmolVLM 256M0.256B
SMOLLM·8K CTX· CHAT· VISION
91
TOK/S · 32% VRAM
S
Gemma 3 270M0.27B
GEMMA·32K CTX· CHAT
86
TOK/S · 33% VRAM
S
Snowflake Arctic Embed M v2.00.305B
EMBEDDING·8K CTX· EMBEDDING· MULTILINGUAL
76
TOK/S · 34% VRAM
S
bge-large-en-v1.5 335M0.335B
EMBEDDING·1K CTX· CHAT
69
TOK/S · 35% VRAM
S
mxbai-embed-large-v10.335B
EMBEDDING·1K CTX· EMBEDDING
69
TOK/S · 35% VRAM
S
Snowflake Arctic Embed L0.335B
EMBEDDING·1K CTX· EMBEDDING
69
TOK/S · 35% VRAM
S
LFM2 350M0.35B
LFM·125K CTX· CHAT· TOOL_USE
66
TOK/S · 35% VRAM
S
SmolLM2 360M0.36B
SMOLLM·8K CTX· CHAT
64
TOK/S · 35% VRAM
S
GPT-2 Medium 345M0.38B
GPT2·1K CTX· CHAT
61
TOK/S · 36% VRAM
A
Qwen 2.5 0.5B0.5B
QWEN·32K CTX· CHAT
46
TOK/S · 40% VRAM
A
SmolVLM 500M0.5B
SMOLLM·8K CTX· CHAT· VISION
46
TOK/S · 40% VRAM
A
Falcon-H1 0.5B0.52B
FALCON·128K CTX· CHAT
45
TOK/S · 40% VRAM
A
BGE-M30.568B
EMBEDDING·8K CTX· EMBEDDING
41
TOK/S · 42% VRAM
A
Snowflake Arctic Embed L v2.00.568B
EMBEDDING·8K CTX· EMBEDDING· MULTILINGUAL
41
TOK/S · 42% VRAM
B
Qwen3-Embedding 0.6B0.6B
EMBEDDING·32K CTX· EMBEDDING· MULTILINGUAL
39
TOK/S · 43% VRAM
B
Falcon-H1R Tiny 0.6B0.6B
FALCON·32K CTX· CHAT· REASONING
39
TOK/S · 43% VRAM
B
Falcon Perception 0.6B0.6B
FALCON·4K CTX· VISION
39
TOK/S · 43% VRAM
B
Qwen 1.5 0.5B0.62B
QWEN·32K CTX· CHAT
37
TOK/S · 43% VRAM
B
Qwen3 0.6B0.75B
QWEN·32K CTX· CHAT· REASONING
31
TOK/S · 47% VRAM
B
LFM2 700M0.74B
LFM·125K CTX· CHAT· TOOL_USE
31
TOK/S · 47% VRAM
B
GPT-2 Large 774M0.81B
GPT2·1K CTX· CHAT
29
TOK/S · 49% VRAM
B
Qwen 3.5 0.8B0.87B
QWEN·256K CTX· CHAT· CODING· MULTILINGUAL
27
TOK/S · 51% VRAM
B
InternVL3 1B0.94B
OTHER·32K CTX· CHAT· VISION
25
TOK/S · 53% VRAM
C
TinyLlama 1.1B1.1B
LLAMA·2K CTX· CHAT
21
TOK/S · 58% VRAM
C
MiniCPM5 1B1.08B
MINICPM·128K CTX· CHAT· REASONING· MULTILINGUAL
21
TOK/S · 57% VRAM
C
LFM2.5-1.2B-Thinking1.2B
LFM·122K CTX· CHAT· REASONING· TOOL_USE
19
TOK/S · 61% VRAM
C
Llama-3.2-1B1.2B
LLAMA·4K CTX· CHAT
19
TOK/S · 61% VRAM
C
LFM2 1.2B1.2B
LFM·125K CTX· CHAT· TOOL_USE· MULTILINGUAL
19
TOK/S · 61% VRAM
C
Zamba2 1.2B1.2B
OTHER·4K CTX· CHAT
19
TOK/S · 61% VRAM
C
EXAONE-4.0-1.2B1.3B
EXAONE·64K CTX· CHAT
18
TOK/S · 64% VRAM
C
OPT 1.3B1.3B
OPT·2K CTX· CHAT
18
TOK/S · 64% VRAM
C
MiniCPM-V 4.61.3B
OTHER·256K CTX· CHAT· VISION
18
TOK/S · 64% VRAM
C
DeepSeek Coder 1.3B1.35B
DEEPSEEK·16K CTX· CODING
17
TOK/S · 66% VRAM
C
Phi-1 1.3B1.42B
PHI·2K CTX· CODING
16
TOK/S · 68% VRAM
C
Phi-1.5 1.3B1.42B
PHI·2K CTX· CHAT· CODING
16
TOK/S · 68% VRAM
C
stablelm-2-1_6b1.6B
STABLELM·4K CTX· CHAT
15
TOK/S · 73% VRAM
▸ NEXT STEP

Get personalized recommendations.

See ranked models with benchmark scores, run commands, and precise speed estimates for your NVS 510.

WHAT THIS CARD IS WORTH

NVS 510 holds 47 of the models in our catalogue and is, in practice, a Q4_K_M card — the largest it takes is Moondream2 1.9B at Q4_K_M.

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
Moondream2 1.9B1.9B · Q4_K_M1.7 GB19 tok/sEST
Falcon3-1B1.67B · Q5_K_M1.8 GB18 tok/sEST
LFM2-VL 1.6B1.6B · Q5_K_M1.7 GB19 tok/sEST
Falcon-H1 1.5B1.55B · Q6_K1.7 GB17 tok/sEST
Qwen2.5-Coder-1.5B1.5B · Q6_K1.6 GB18 tok/sEST
Qwen2 Math 1.5B1.5B · Q6_K1.6 GB18 tok/sEST
Qwen 2.5 1.5B1.5B · Q6_K1.6 GB18 tok/sEST
Stella en 1.5B v51.5B · Q6_K1.6 GB18 tok/sEST
AGAINST CARDS OF SIMILAR MEMORY
M2 (8GB)5 GB$5992.8 tok/s per $100
M3 (8GB)5 GB$5992.8 tok/s per $100
M1 (8GB)5 GB$4992.2 tok/s per $100
RTX 2060 6GB6 GB$15026.7 tok/s per $100
NVS 5102 GB

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 NVS 510 — 2 GB VRAM.

NVS 510 SPEC
BRAND
NVIDIA
VRAM
2 GB DDR3
BANDWIDTH
29 GB/s
FP16 COMPUTE
0.3 TFLOPS
FP32 COMPUTE
0.3 TFLOPS
CUDA CORES
192
TDP
35 W
ARCHITECTURE
Kepler
▸ AI CAPABILITY
47/ 449 models @ Q4

With 2 GB VRAM and 29 GB/s bandwidth, this GPU handles models up to 1.61B parameters.

Speed ≈ bandwidth / model_size × efficiency. A 7B model at Q4 runs at ~3 tok/s.

§ 01TOP MODELS FOR NVS 510
47 FIT · SHOWING 20
MODELSIZEVRAM Q4TOK/SAVG
GPT-2 XL 1.5B1.61B1.5 GB145.1
stablelm-2-1_6b1.6B1.5 GB159.5
Falcon-H1 1.5B1.55B1.4 GB1543.8
Qwen2.5-Coder-1.5B1.5B1.4 GB1519.6
Qwen2 Math 1.5B1.5B1.4 GB1519.6
Qwen 2.5 1.5B1.5B1.4 GB1530.2
Yi Coder 1.5B1.5B1.4 GB1514.6
Stella en 1.5B v51.5B1.4 GB15
Phi-1 1.3B1.42B1.4 GB167.2
Phi-1.5 1.3B1.42B1.4 GB167.2
DeepSeek Coder 1.3B1.35B1.3 GB1716.8
EXAONE-4.0-1.2B1.3B1.3 GB1818.9
OPT 1.3B1.3B1.3 GB185.3
MiniCPM-V 4.61.3B1.3 GB1821.5
LFM2.5-1.2B-Thinking1.2B1.2 GB1919.6
Llama-3.2-1B1.2B1.2 GB1910.1
LFM2 1.2B1.2B1.2 GB1915.6
Zamba2 1.2B1.2B1.2 GB1941.5
TinyLlama 1.1B1.1B1.2 GB2113.6
MiniCPM5 1B1.08B1.1 GB2126.5