ARM· 5TH GEN

ARM Mali-G625 MC2

VRAM
3 GB
BUDGET
BANDWIDTH
14.9
GB/S
MODELS Q4
88/428
21%
7B Q4 SPEED
~2
SLOW
▸ MODEL COVERAGE @ Q421% OF ALL
▸ ESTIMATED SPEED· BY MODEL SIZE @ Q4

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

3B
~4
TOK/S
7B
3.9GB NEEDED
14B
7.9GB NEEDED
32B
18.0GB NEEDED
70B
39.4GB NEEDED
▸ SPECIFICATIONS
VRAM
3 GB
BANDWIDTH
14.9 GB/s
FP16 COMPUTE
0.15 TFLOPS
TDP
2W
MEMORY
ARCHITECTURE
5th Gen
13
FAST MODELS · >30 TOK/S
Real-time chat speed
41
USABLE · >10 TOK/S
Comfortable for all tasks
88
TOTAL COMPATIBLE
Fit in VRAM at Q4
▸ DON’T WANT TO BUY?

Test ARM Mali-G625 MC2 (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· 88
S
Falcon-H1R Tiny 90M0.09B
FALCON·256K CTX· CHAT· REASONING
147
TOK/S · 18% VRAM
S
nomic-embed-text-v1.5 100M0.1B
EMBEDDING·8K CTX· CHAT
132
TOK/S · 18% VRAM
S
GPT-2 124M0.124B
GPT2·1K CTX· CHAT
107
TOK/S · 19% VRAM
S
SmolLM2 135M0.135B
SMOLLM·2K CTX· CHAT
98
TOK/S · 19% VRAM
A
SmolVLM 256M0.256B
SMOLLM·8K CTX· CHAT· VISION
52
TOK/S · 21% VRAM
A
Gemma 3 270M0.27B
GEMMA·32K CTX· CHAT
49
TOK/S · 22% VRAM
A
Snowflake Arctic Embed M v2.00.305B
EMBEDDING·8K CTX· EMBEDDING· MULTILINGUAL
43
TOK/S · 22% VRAM
A
bge-large-en-v1.5 335M0.335B
EMBEDDING·1K CTX· CHAT
40
TOK/S · 23% VRAM
A
mxbai-embed-large-v10.335B
EMBEDDING·1K CTX· EMBEDDING
40
TOK/S · 23% VRAM
A
Snowflake Arctic Embed L0.335B
EMBEDDING·1K CTX· EMBEDDING
40
TOK/S · 23% VRAM
B
LFM2 350M0.35B
LFM·125K CTX· CHAT· TOOL_USE
38
TOK/S · 23% VRAM
B
GPT-2 Medium 345M0.345B
GPT2·1K CTX· CHAT
38
TOK/S · 23% VRAM
B
SmolLM2 360M0.36B
SMOLLM·8K CTX· CHAT
37
TOK/S · 24% VRAM
B
Falcon-H1 0.5B0.5B
FALCON·128K CTX· CHAT
26
TOK/S · 26% VRAM
B
Qwen 1.5 0.5B0.5B
QWEN·32K CTX· CHAT
26
TOK/S · 26% VRAM
B
Qwen 2.5 0.5B0.5B
QWEN·32K CTX· CHAT
26
TOK/S · 26% VRAM
B
SmolVLM 500M0.5B
SMOLLM·8K CTX· CHAT· VISION
26
TOK/S · 26% VRAM
C
BGE-M30.568B
EMBEDDING·8K CTX· EMBEDDING
23
TOK/S · 28% VRAM
C
Snowflake Arctic Embed L v2.00.568B
EMBEDDING·8K CTX· EMBEDDING· MULTILINGUAL
23
TOK/S · 28% VRAM
C
Qwen3 0.6B0.6B
QWEN·32K CTX· CHAT· REASONING
22
TOK/S · 29% VRAM
C
Qwen3-Embedding 0.6B0.6B
EMBEDDING·32K CTX· EMBEDDING· MULTILINGUAL
22
TOK/S · 29% VRAM
C
Falcon-H1R Tiny 0.6B0.6B
FALCON·32K CTX· CHAT· REASONING
22
TOK/S · 29% VRAM
C
Falcon Perception 0.6B0.6B
FALCON·4K CTX· VISION
22
TOK/S · 29% VRAM
C
LFM2 700M0.7B
LFM·125K CTX· CHAT· TOOL_USE
19
TOK/S · 31% VRAM
C
Qwen 3.5 0.8B0.8B
QWEN·256K CTX· CHAT· CODING· MULTILINGUAL
17
TOK/S · 33% VRAM
C
GPT-2 Large 774M0.774B
GPT2·1K CTX· CHAT
17
TOK/S · 32% VRAM
C
Qwen3.5-0.8B0.9B
QWEN·256K CTX· CHAT
15
TOK/S · 35% VRAM
D
Falcon3-1B1B
FALCON·32K CTX· CHAT
13
TOK/S · 37% VRAM
D
InternLM2 1B1B
INTERNLM·32K CTX· CHAT
13
TOK/S · 37% VRAM
D
InternVL3 1B1B
OTHER·32K CTX· CHAT· VISION
13
TOK/S · 37% VRAM
D
TinyLlama 1.1B1.1B
LLAMA·2K CTX· CHAT
12
TOK/S · 39% VRAM
D
LFM2.5-1.2B-Thinking1.2B
LFM·122K CTX· CHAT· REASONING· TOOL_USE
11
TOK/S · 41% VRAM
D
Llama-3.2-1B1.2B
LLAMA·4K CTX· CHAT
11
TOK/S · 41% VRAM
D
LFM2 1.2B1.2B
LFM·125K CTX· CHAT· TOOL_USE· MULTILINGUAL
11
TOK/S · 41% VRAM
D
Zamba2 1.2B1.2B
OTHER·4K CTX· CHAT
11
TOK/S · 41% VRAM
D
DeepSeek Coder 1.3B1.3B
DEEPSEEK·16K CTX· CODING
10
TOK/S · 43% VRAM
D
EXAONE-4.0-1.2B1.3B
EXAONE·64K CTX· CHAT
10
TOK/S · 43% VRAM
D
OPT 1.3B1.3B
OPT·2K CTX· CHAT
10
TOK/S · 43% VRAM
D
Phi-1 1.3B1.3B
PHI·2K CTX· CODING
10
TOK/S · 43% VRAM
D
Phi-1.5 1.3B1.3B
PHI·2K CTX· CHAT· CODING
10
TOK/S · 43% VRAM
▸ NEXT STEP

Get personalized recommendations.

See ranked models with benchmark scores, run commands, and precise speed estimates for your ARM Mali-G625 MC2.

▸ DEVICE UNDER TEST

ARM Mali-G625 MC23 GB VRAM.

ARM MALI-G625 MC2 SPEC
BRAND
Apple
VRAM
3 GB
BANDWIDTH
14.9 GB/s
FP16 COMPUTE
0.15 TFLOPS
FP32 COMPUTE
0.07 TFLOPS
TDP
2 W
ARCHITECTURE
5th Gen
▸ AI CAPABILITY
88/ 428 models @ Q4

With 3 GB VRAM and 14.9 GB/s bandwidth, this GPU handles models up to 3.2B parameters.

Speed ≈ bandwidth / model_size × efficiency. A 7B model at Q4 runs at ~2 tok/s.

§ 01TOP MODELS FOR ARM MALI-G625 MC2
88 FIT · SHOWING 20
MODELSIZEVRAM Q4TOK/SAVG
granite-4.0-h-micro 3.2B3.2B2.4 GB418.4
Llama-3.2-3B3.2B2.4 GB417.9
Falcon3-3B3.1B2.4 GB425.7
Qwen 2.5 3B3.1B2.4 GB437.2
SmolLM3-3B3.1B2.4 GB430.5
Cogito 3B3B2.3 GB422.1
Falcon-H1 3B3B2.3 GB449.5
Ministral 3B3B2.3 GB429.6
StarCoder2 3B3B2.3 GB49.5
Granite 4.1 3B3B2.3 GB416.6
Granite 4.0 3B Vision3B2.3 GB423.9
xLAM-2 3B Function-Calling3B2.3 GB4
Jamba 2 3B3B2.3 GB4
Dolly v2 3B2.8B2.2 GB55.6
StableLM Zephyr 3B2.79B2.2 GB514.9
Zephyr 3B2.79B2.2 GB514.4
OPT 2.7B2.7B2.1 GB528.0
Phi-2 2.7B2.7B2.1 GB524.1
Zamba2 2.7B2.7B2.1 GB548.0
gemma-2-2b2.6B2.1 GB522.9