FitMyLLM
▸ APPLE· M4 PRO

Apple M4 Pro (48GB)

Running LLMs on the Apple M4 Pro (48GB) — the long read: which models fit at which quantisation, and the settings worth changing. · Or what else $1,799 buys

USABLE MEMORY
36 GB
HIGH-END
BANDWIDTH
273
GB/S
MODELS Q4
342/449
76%
7B Q4 SPEED
~35
FAST
▸ MODEL COVERAGE @ Q476% OF ALL
▸ ESTIMATED SPEED· BY MODEL SIZE @ Q4

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

3B
~81
TOK/S
7B
~35
TOK/S
14B
~17
TOK/S
32B
~8
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
USABLE MEMORY
36 GB of 48
BANDWIDTH
273 GB/s
FP16 COMPUTE
9.2 TFLOPS
TDP
30W
MEMORY
Unified
ARCHITECTURE
M4 Pro
MSRP
$1,799
217
FAST MODELS · >30 TOK/S
Real-time chat speed
295
USABLE · >10 TOK/S
Comfortable for all tasks
342
TOTAL COMPATIBLE
Fit in usable memory at Q4
▸ DON’T WANT TO BUY?

Test M4 Pro (48GB) (or anything bigger) without committing. Pay by the second, cancel anytime.

Rent on Vast.ai →Or RunPod →

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· 342
S
Falcon-H1R Tiny 90M0.09B
FALCON·256K CTX· CHAT· REASONING
2696
TOK/S · 2% MEM
›
S
SmolLM2 135M0.135B
SMOLLM·2K CTX· CHAT
1798
TOK/S · 2% MEM
›
S
nomic-embed-text-v1.5 100M0.14B
EMBEDDING·8K CTX· CHAT
1733
TOK/S · 2% MEM
›
S
GPT-2 124M0.14B
GPT2·1K CTX· CHAT
1733
TOK/S · 2% MEM
›
S
SmolVLM 256M0.256B
SMOLLM·8K CTX· CHAT· VISION
948
TOK/S · 2% MEM
›
S
Gemma 3 270M0.27B
GEMMA·32K CTX· CHAT
899
TOK/S · 2% MEM
›
S
Snowflake Arctic Embed M v2.00.305B
EMBEDDING·8K CTX· EMBEDDING· MULTILINGUAL
796
TOK/S · 2% MEM
›
S
bge-large-en-v1.5 335M0.335B
EMBEDDING·1K CTX· CHAT
724
TOK/S · 2% MEM
›
S
mxbai-embed-large-v10.335B
EMBEDDING·1K CTX· EMBEDDING
724
TOK/S · 2% MEM
›
S
Snowflake Arctic Embed L0.335B
EMBEDDING·1K CTX· EMBEDDING
724
TOK/S · 2% MEM
›
S
LFM2 350M0.35B
LFM·125K CTX· CHAT· TOOL_USE
693
TOK/S · 2% MEM
›
S
SmolLM2 360M0.36B
SMOLLM·8K CTX· CHAT
674
TOK/S · 2% MEM
›
S
GPT-2 Medium 345M0.38B
GPT2·1K CTX· CHAT
639
TOK/S · 2% MEM
›
S
Qwen 2.5 0.5B0.5B
QWEN·32K CTX· CHAT
485
TOK/S · 2% MEM
›
S
SmolVLM 500M0.5B
SMOLLM·8K CTX· CHAT· VISION
485
TOK/S · 2% MEM
›
S
Falcon-H1 0.5B0.52B
FALCON·128K CTX· CHAT
467
TOK/S · 2% MEM
›
S
BGE-M30.568B
EMBEDDING·8K CTX· EMBEDDING
427
TOK/S · 2% MEM
›
S
Snowflake Arctic Embed L v2.00.568B
EMBEDDING·8K CTX· EMBEDDING· MULTILINGUAL
427
TOK/S · 2% MEM
›
S
Qwen3-Embedding 0.6B0.6B
EMBEDDING·32K CTX· EMBEDDING· MULTILINGUAL
404
TOK/S · 2% MEM
›
S
Falcon-H1R Tiny 0.6B0.6B
FALCON·32K CTX· CHAT· REASONING
404
TOK/S · 2% MEM
›
S
Falcon Perception 0.6B0.6B
FALCON·4K CTX· VISION
404
TOK/S · 2% MEM
›
S
Qwen 1.5 0.5B0.62B
QWEN·32K CTX· CHAT
391
TOK/S · 2% MEM
›
S
LFM2 700M0.74B
LFM·125K CTX· CHAT· TOOL_USE
328
TOK/S · 3% MEM
›
S
Qwen3 0.6B0.75B
QWEN·32K CTX· CHAT· REASONING
324
TOK/S · 3% MEM
›
S
GPT-2 Large 774M0.81B
GPT2·1K CTX· CHAT
300
TOK/S · 3% MEM
›
S
Qwen 3.5 0.8B0.87B
QWEN·256K CTX· CHAT· CODING· MULTILINGUAL
279
TOK/S · 3% MEM
›
S
InternVL3 1B0.94B
OTHER·32K CTX· CHAT· VISION
258
TOK/S · 3% MEM
›
S
MiniCPM5 1B1.08B
MINICPM·128K CTX· CHAT· REASONING· MULTILINGUAL
225
TOK/S · 3% MEM
›
S
TinyLlama 1.1B1.1B
LLAMA·2K CTX· CHAT
221
TOK/S · 3% MEM
›
S
LFM2.5-1.2B-Thinking1.2B
LFM·122K CTX· CHAT· REASONING· TOOL_USE
202
TOK/S · 3% MEM
›
S
Llama-3.2-1B1.2B
LLAMA·4K CTX· CHAT
202
TOK/S · 3% MEM
›
S
LFM2 1.2B1.2B
LFM·125K CTX· CHAT· TOOL_USE· MULTILINGUAL
202
TOK/S · 3% MEM
›
S
Zamba2 1.2B1.2B
OTHER·4K CTX· CHAT
202
TOK/S · 3% MEM
›
S
EXAONE-4.0-1.2B1.3B
EXAONE·64K CTX· CHAT
187
TOK/S · 4% MEM
›
S
OPT 1.3B1.3B
OPT·2K CTX· CHAT
187
TOK/S · 4% MEM
›
S
MiniCPM-V 4.61.3B
OTHER·256K CTX· CHAT· VISION
187
TOK/S · 4% MEM
›
S
DeepSeek Coder 1.3B1.35B
DEEPSEEK·16K CTX· CODING
180
TOK/S · 4% MEM
›
S
Phi-1 1.3B1.42B
PHI·2K CTX· CODING
171
TOK/S · 4% MEM
›
S
Phi-1.5 1.3B1.42B
PHI·2K CTX· CHAT· CODING
171
TOK/S · 4% MEM
›
S
LFM2 8B A1B8.3BMoE
LFM·125K CTX· CHAT· TOOL_USE· MULTILINGUAL
162
TOK/S · 15% MEM
›
▸ NEXT STEP

Get personalized recommendations.

See ranked models with benchmark scores, run commands, and precise speed estimates for your Apple M4 Pro (48GB).

▸Find models for Apple M4 Pro (48GB)Compare with other GPUs
WHAT THIS CARD IS WORTH

M4 Pro (48GB) holds 342 of the models in our catalogue and is, in practice, a IQ4_XS card — the largest it takes is Jamba 2 Mini at IQ4_XS.

TOKENS/SEC PER $100
1.9
8B at Q4_K_M, so cards compare like for like
USABLE MEMORY PER $100
2.00 GB
what memory costs on this card
THE BIGGEST IT TAKES
Jamba 2 Mini52B · IQ4_XS32.0 GB24 tok/sEST
Jamba 1.5 Mini 52B51.6B · IQ4_XS31.8 GB24 tok/sEST
Kimi-Linear-48B-A3B48B · Q4_K_M32.0 GB86 tok/sEST
Nemotron-H 47B47B · Q4_K_S31.5 GB6 tok/sEST
Mixtral-8x7B46.7B · Q4_K_M31.5 GB20 tok/sEST
Nous-Hermes-2-Mixtral-8x7B-DPO46.7B · Q4_K_M31.5 GB20 tok/sEST
Dolphin 2.6 Mixtral 8x7B46.7B · Q4_K_M31.5 GB20 tok/sEST
Phi-3.5 MoE 42B41.9B · Q5_K_S32.2 GB34 tok/sEST
AGAINST CARDS OF SIMILAR MEMORY
M1 Ultra (64GB)48 GB$2,4992.9 tok/s per $100
M2 Ultra (64GB)48 GB$2,9992.4 tok/s per $100
M4 Max (64GB)48 GB$2,8991.9 tok/s per $100
M2 Max (64GB)48 GB$2,2992.0 tok/s per $100
M4 Pro (48GB)36 GB$1,7991.9 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

Apple M4 Pro (48GB) — 48 GB unified, 36 GB usable.

▸ APPLE M4 PRO (48GB) SPEC
BRAND
Apple
UNIFIED MEMORY
48 GB
USABLE BY A MODEL
36 GB
BANDWIDTH
273 GB/s
FP16 COMPUTE
9.2 TFLOPS
FP32 COMPUTE
9.2 TFLOPS
TDP
30 W
ARCHITECTURE
M4 Pro
MSRP
$1799
▸ AI CAPABILITY
342/ 449 models @ Q4

With 36 GB of its 48 GB reaching a model and 273 GB/s bandwidth, this machine handles models up to 48B parameters.

Speed ≈ bandwidth / model_size × efficiency. A 7B model at Q4 runs at ~31 tok/s.

RENT IT — LIVE PRICES
Checking what the clouds are charging…
§ 01TOP MODELS FOR APPLE M4 PRO (48GB)
342 FIT · SHOWING 20
MODELSIZEVRAM Q4TOK/SAVG
Kimi-Linear-48B-A3B48B29.8 GB8126.6
Nemotron-H 47B47B29.2 GB584.6
Mixtral-8x7B46.7B29.0 GB1918.8
Nous-Hermes-2-Mixtral-8x7B-DPO46.7B29.0 GB1927.4
Dolphin 2.6 Mixtral 8x7B46.7B29.0 GB1923.8
Phi-3.5 MoE 42B41.9B26.1 GB3756.7
Falcon 40B40B24.9 GB620.9
InternVL3 38B38B23.7 GB678.9
Seed-OSS 36B Instruct36B22.5 GB754.4
c4ai-command-r-v01 35B35B21.9 GB727.5
Qwen 3.5 35B A3B35B21.9 GB8153.3
Qwen 3.6 35B A3B35B21.9 GB8153.9
Nous Capybara 34B34.4B21.5 GB742.0
Yi-1.5 34B34.4B21.5 GB745.3
Falcon-H1 34B34B21.3 GB766.1
CodeLlama 34B34B21.3 GB725.4
Nous Hermes 2 34B34B21.3 GB747.0
Phind CodeLlama 34B34B21.3 GB768.1
LLaVA-1.6 Yi 34B34B21.3 GB747.4
WizardCoder Python 34B34B21.3 GB773.2