FitMyLLM
▸ INTEL· GENERATION 12.5

Intel Arctic Sound-M

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

VRAM
16 GB
MID-RANGE
BANDWIDTH
1230
GB/S
MODELS Q4
262/449
58%
7B Q4 SPEED
~56
BLAZING
▸ MODEL COVERAGE @ Q458% OF ALL
▸ ESTIMATED SPEED· BY MODEL SIZE @ Q4

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

3B
~131
TOK/S
7B
~56
TOK/S
14B
~28
TOK/S
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
16 GB
BANDWIDTH
1230 GB/s
FP16 COMPUTE
29.5 TFLOPS
TDP
500W
MEMORY
HBM2e
ARCHITECTURE
Generation 12.5
COMPUTE UNITS
512
PCIE
Gen 4 x16
244
FAST MODELS · >30 TOK/S
Real-time chat speed
262
USABLE · >10 TOK/S
Comfortable for all tasks
262
TOTAL COMPATIBLE
Fit in VRAM at Q4
▸ DON’T WANT TO BUY?

Test Arctic Sound-M (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· 262
S
Falcon-H1R Tiny 90M0.09B
FALCON·256K CTX· CHAT· REASONING
4373
TOK/S · 3% VRAM
›
S
SmolLM2 135M0.135B
SMOLLM·2K CTX· CHAT
2916
TOK/S · 4% VRAM
›
S
nomic-embed-text-v1.5 100M0.14B
EMBEDDING·8K CTX· CHAT
2811
TOK/S · 4% VRAM
›
S
GPT-2 124M0.14B
GPT2·1K CTX· CHAT
2811
TOK/S · 4% VRAM
›
S
SmolVLM 256M0.256B
SMOLLM·8K CTX· CHAT· VISION
1537
TOK/S · 4% VRAM
›
S
Gemma 3 270M0.27B
GEMMA·32K CTX· CHAT
1458
TOK/S · 4% VRAM
›
S
Snowflake Arctic Embed M v2.00.305B
EMBEDDING·8K CTX· EMBEDDING· MULTILINGUAL
1290
TOK/S · 4% VRAM
›
S
bge-large-en-v1.5 335M0.335B
EMBEDDING·1K CTX· CHAT
1175
TOK/S · 4% VRAM
›
S
mxbai-embed-large-v10.335B
EMBEDDING·1K CTX· EMBEDDING
1175
TOK/S · 4% VRAM
›
S
Snowflake Arctic Embed L0.335B
EMBEDDING·1K CTX· EMBEDDING
1175
TOK/S · 4% VRAM
›
S
LFM2 350M0.35B
LFM·125K CTX· CHAT· TOOL_USE
1125
TOK/S · 4% VRAM
›
S
SmolLM2 360M0.36B
SMOLLM·8K CTX· CHAT
1093
TOK/S · 4% VRAM
›
S
GPT-2 Medium 345M0.38B
GPT2·1K CTX· CHAT
1036
TOK/S · 5% VRAM
›
S
Qwen 2.5 0.5B0.5B
QWEN·32K CTX· CHAT
787
TOK/S · 5% VRAM
›
S
SmolVLM 500M0.5B
SMOLLM·8K CTX· CHAT· VISION
787
TOK/S · 5% VRAM
›
S
Falcon-H1 0.5B0.52B
FALCON·128K CTX· CHAT
757
TOK/S · 5% VRAM
›
S
BGE-M30.568B
EMBEDDING·8K CTX· EMBEDDING
693
TOK/S · 5% VRAM
›
S
Snowflake Arctic Embed L v2.00.568B
EMBEDDING·8K CTX· EMBEDDING· MULTILINGUAL
693
TOK/S · 5% VRAM
›
S
Qwen3-Embedding 0.6B0.6B
EMBEDDING·32K CTX· EMBEDDING· MULTILINGUAL
656
TOK/S · 5% VRAM
›
S
Falcon-H1R Tiny 0.6B0.6B
FALCON·32K CTX· CHAT· REASONING
656
TOK/S · 5% VRAM
›
S
Falcon Perception 0.6B0.6B
FALCON·4K CTX· VISION
656
TOK/S · 5% VRAM
›
S
Qwen 1.5 0.5B0.62B
QWEN·32K CTX· CHAT
635
TOK/S · 5% VRAM
›
S
LFM2 700M0.74B
LFM·125K CTX· CHAT· TOOL_USE
532
TOK/S · 6% VRAM
›
S
Qwen3 0.6B0.75B
QWEN·32K CTX· CHAT· REASONING
525
TOK/S · 6% VRAM
›
S
GPT-2 Large 774M0.81B
GPT2·1K CTX· CHAT
486
TOK/S · 6% VRAM
›
S
Qwen 3.5 0.8B0.87B
QWEN·256K CTX· CHAT· CODING· MULTILINGUAL
452
TOK/S · 6% VRAM
›
S
InternVL3 1B0.94B
OTHER·32K CTX· CHAT· VISION
419
TOK/S · 7% VRAM
›
S
MiniCPM5 1B1.08B
MINICPM·128K CTX· CHAT· REASONING· MULTILINGUAL
364
TOK/S · 7% VRAM
›
S
TinyLlama 1.1B1.1B
LLAMA·2K CTX· CHAT
358
TOK/S · 7% VRAM
›
S
LFM2.5-1.2B-Thinking1.2B
LFM·122K CTX· CHAT· REASONING· TOOL_USE
328
TOK/S · 8% VRAM
›
S
Llama-3.2-1B1.2B
LLAMA·4K CTX· CHAT
328
TOK/S · 8% VRAM
›
S
LFM2 1.2B1.2B
LFM·125K CTX· CHAT· TOOL_USE· MULTILINGUAL
328
TOK/S · 8% VRAM
›
S
Zamba2 1.2B1.2B
OTHER·4K CTX· CHAT
328
TOK/S · 8% VRAM
›
S
EXAONE-4.0-1.2B1.3B
EXAONE·64K CTX· CHAT
303
TOK/S · 8% VRAM
›
S
OPT 1.3B1.3B
OPT·2K CTX· CHAT
303
TOK/S · 8% VRAM
›
S
MiniCPM-V 4.61.3B
OTHER·256K CTX· CHAT· VISION
303
TOK/S · 8% VRAM
›
S
DeepSeek Coder 1.3B1.35B
DEEPSEEK·16K CTX· CODING
292
TOK/S · 8% VRAM
›
S
Phi-1 1.3B1.42B
PHI·2K CTX· CODING
277
TOK/S · 8% VRAM
›
S
Phi-1.5 1.3B1.42B
PHI·2K CTX· CHAT· CODING
277
TOK/S · 8% VRAM
›
S
LFM2 8B A1B8.3BMoE
LFM·125K CTX· CHAT· TOOL_USE· MULTILINGUAL
262
TOK/S · 35% VRAM
›
▸ NEXT STEP

Get personalized recommendations.

See ranked models with benchmark scores, run commands, and precise speed estimates for your Arctic Sound-M.

WHAT THIS CARD IS WORTH

Arctic Sound-M holds 262 of the models in our catalogue and is, in practice, a IQ4_XS card — the largest it takes is ERNIE 4.5 21B A3B 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
ERNIE 4.5 21B A3B21.95B · IQ4_XS13.9 GB132 tok/sEST
GPT-OSS 20B21B · Q4_K_S13.8 GB105 tok/sEST
Reka Flash 321B · Q4_K_S14.2 GB18 tok/sEST
Reka Flash 3.121B · Q4_K_S14.2 GB18 tok/sEST
InternLM2 20B19.8B · Q4_K_M14.3 GB18 tok/sEST
InternLM2.5 20B19.8B · Q4_K_M14.3 GB18 tok/sEST
Ling-lite 16.8B16.8B · Q5_K_M13.6 GB129 tok/sEST
DeepSeek V2 Lite 16B16B · Q5_K_M12.9 GB129 tok/sEST
AGAINST CARDS OF SIMILAR MEMORY
M1 Max (32GB)21 GB$1,4993.0 tok/s per $100
M2 Max (32GB)21 GB$1,7992.5 tok/s per $100
M2 Pro (32GB)21 GB$1,4991.9 tok/s per $100
M4 (32GB)21 GB$1,1991.7 tok/s per $100
Arctic Sound-M16 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

Intel Arctic Sound-M — 16 GB VRAM.

▸ ARCTIC SOUND-M SPEC
BRAND
Intel
VRAM
16 GB HBM2e
BANDWIDTH
1230 GB/s
FP16 COMPUTE
29.5 TFLOPS
FP32 COMPUTE
14.8 TFLOPS
TDP
500 W
ARCHITECTURE
Generation 12.5
▸ AI CAPABILITY
262/ 449 models @ Q4

With 16 GB VRAM and 1230 GB/s bandwidth, this GPU handles models up to 19.8B parameters.

Speed ≈ bandwidth / model_size × efficiency. A 7B model at Q4 runs at ~141 tok/s.

RENT IT — LIVE PRICES
Checking what the clouds are charging…
§ 01TOP MODELS FOR ARCTIC SOUND-M
262 FIT · SHOWING 20
MODELSIZEVRAM Q4TOK/SAVG
InternLM2 20B19.8B12.6 GB2045.1
InternLM2.5 20B19.8B12.6 GB2050.9
Ling-lite 16.8B16.8B10.8 GB164—
DeepSeek V2 Lite 16B16B10.3 GB16438.0
StarCoder2 15B15.96B10.2 GB2526.5
DeepSeek-Coder-V2-Lite 15.7B15.7B10.1 GB16443.0
DeepSeek-VL2 Small 16B15.7B10.1 GB16443.1
StarCoder 15B15.5B10.0 GB2521.0
InternVL3 14B15.12B9.7 GB2638.1
Phi-4-reasoning-vision 15B15B9.7 GB2642.8
DeepSeek R1 Distill Qwen 14B14.8B9.5 GB2743.9
DeepCoder 14B14.8B9.5 GB2738.7
Qwen2.5-Coder-14B14.8B9.5 GB2741.3
Qwen2.5-14B14.8B9.5 GB2741.3
Qwen3 14B14.8B9.5 GB2745.7
phi-4 14B14.66B9.4 GB2733.7
Phi-4-reasoning 14B14.66B9.4 GB2733.7
Phi-4-reasoning-plus 14B14.66B9.4 GB2775.5
Ministral 3 14B14B9.0 GB2825.9
Phi-3-medium-14b14B9.0 GB2833.7