AMD· GCN 1.0

AMD Radeon R7 350 640SP

VRAM
2 GB
BUDGET
BANDWIDTH
72
GB/S
MODELS Q4
42/331
13%
7B Q4 SPEED
~9
USABLE
▸ MODEL COVERAGE @ Q413% OF ALL
▸ ESTIMATED SPEED· BY MODEL SIZE @ Q4

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

3B
~21
TOK/S
7B
3.9GB NEEDED
14B
7.9GB NEEDED
32B
18.0GB NEEDED
70B
39.4GB NEEDED
▸ SPECIFICATIONS
VRAM
2 GB
BANDWIDTH
72 GB/s
FP16 COMPUTE
1.2 TFLOPS
TDP
55W
MEMORY
GDDR5
ARCHITECTURE
GCN 1.0
STREAM PROCESSORS
640
COMPUTE UNITS
10
PCIE
Gen 3 x16
42
FAST MODELS · >30 TOK/S
Real-time chat speed
42
USABLE · >10 TOK/S
Comfortable for all tasks
42
TOTAL COMPATIBLE
Fit in VRAM at Q4
▸ DON’T WANT TO BUY?

Test Radeon R7 350 640SP (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· 42
S
nomic-embed-text-v1.5 100M0.1B
EMBEDDING·8K CTX· CHAT
640
TOK/S · 27% VRAM
S
GPT-2 124M0.124B
GPT2·1K CTX· CHAT
516
TOK/S · 28% VRAM
S
SmolLM2 135M0.135B
SMOLLM·2K CTX· CHAT
474
TOK/S · 29% VRAM
S
bge-large-en-v1.5 335M0.335B
EMBEDDING·1K CTX· CHAT
191
TOK/S · 35% VRAM
S
mxbai-embed-large-v10.335B
EMBEDDING·1K CTX· EMBEDDING
191
TOK/S · 35% VRAM
S
Snowflake Arctic Embed L0.335B
EMBEDDING·1K CTX· EMBEDDING
191
TOK/S · 35% VRAM
S
GPT-2 Medium 345M0.345B
GPT2·1K CTX· CHAT
186
TOK/S · 35% VRAM
S
SmolLM2 360M0.36B
SMOLLM·8K CTX· CHAT
178
TOK/S · 35% VRAM
S
Falcon-H1 0.5B0.5B
FALCON·128K CTX· CHAT
128
TOK/S · 40% VRAM
S
Qwen 1.5 0.5B0.5B
QWEN·32K CTX· CHAT
128
TOK/S · 40% VRAM
S
Qwen 2.5 0.5B0.5B
QWEN·32K CTX· CHAT
128
TOK/S · 40% VRAM
S
BGE-M30.568B
EMBEDDING·8K CTX· EMBEDDING
113
TOK/S · 42% VRAM
S
Qwen3 0.6B0.6B
QWEN·32K CTX· CHAT· REASONING
107
TOK/S · 43% VRAM
S
GPT-2 Large 774M0.774B
GPT2·1K CTX· CHAT
83
TOK/S · 48% VRAM
S
Qwen 3.5 0.8B0.8B
QWEN·256K CTX· CHAT· CODING· MULTILINGUAL
80
TOK/S · 49% VRAM
S
Qwen3.5-0.8B0.9B
QWEN·256K CTX· CHAT
71
TOK/S · 52% VRAM
S
Falcon3-1B1B
FALCON·32K CTX· CHAT
64
TOK/S · 55% VRAM
S
InternLM2 1B1B
INTERNLM·32K CTX· CHAT
64
TOK/S · 55% VRAM
A
TinyLlama 1.1B1.1B
LLAMA·2K CTX· CHAT
58
TOK/S · 58% VRAM
A
LFM2.5-1.2B-Thinking1.2B
LFM·122K CTX· CHAT· REASONING· TOOL_USE
53
TOK/S · 61% VRAM
A
Llama-3.2-1B1.2B
LLAMA·4K CTX· CHAT
53
TOK/S · 61% VRAM
A
DeepSeek Coder 1.3B1.3B
DEEPSEEK·16K CTX· CODING
49
TOK/S · 64% VRAM
A
EXAONE-4.0-1.2B1.3B
EXAONE·64K CTX· CHAT
49
TOK/S · 64% VRAM
A
OPT 1.3B1.3B
OPT·2K CTX· CHAT
49
TOK/S · 64% VRAM
A
Phi-1 1.3B1.3B
PHI·2K CTX· CODING
49
TOK/S · 64% VRAM
A
Phi-1.5 1.3B1.3B
PHI·2K CTX· CHAT· CODING
49
TOK/S · 64% VRAM
A
Falcon-H1 1.5B1.5B
FALCON·128K CTX· CHAT· CODING
43
TOK/S · 70% VRAM
A
GPT-2 XL 1.5B1.5B
GPT2·1K CTX· CHAT
43
TOK/S · 70% VRAM
A
Qwen2.5-Coder-1.5B1.5B
QWEN·32K CTX· CHAT· TOOL_USE· CODING
43
TOK/S · 70% VRAM
A
Qwen2 Math 1.5B1.5B
QWEN·4K CTX· REASONING
43
TOK/S · 70% VRAM
A
Qwen 2.5 1.5B1.5B
QWEN·32K CTX· CHAT· CODING
43
TOK/S · 70% VRAM
A
Yi Coder 1.5B1.5B
YI·125K CTX· CODING
43
TOK/S · 70% VRAM
A
stablelm-2-1_6b1.6B
STABLELM·4K CTX· CHAT
40
TOK/S · 73% VRAM
B
Qwen3 1.7B1.7B
QWEN·32K CTX· CHAT· REASONING
38
TOK/S · 76% VRAM
B
SmolLM2 1.7B1.71B
SMOLLM·8K CTX· CHAT
37
TOK/S · 77% VRAM
B
Qwen 1.5 1.8B1.8B
QWEN·32K CTX· CHAT
36
TOK/S · 79% VRAM
B
Moondream2 1.9B1.9B
OTHER·2K CTX· VISION· CHAT
34
TOK/S · 82% VRAM
B
Gemma 1 2B2B
GEMMA·8K CTX· CHAT
32
TOK/S · 86% VRAM
B
Granite 3.0 2B2B
GRANITE·128K CTX· CHAT· CODING
32
TOK/S · 86% VRAM
B
Granite 3.1 2B2B
GRANITE·128K CTX· CHAT· CODING
32
TOK/S · 86% VRAM
▸ NEXT STEP

Get personalized recommendations.

See ranked models with benchmark scores, run commands, and precise speed estimates for your Radeon R7 350 640SP.

▸ DEVICE UNDER TEST

AMD Radeon R7 350 640SP2 GB VRAM.

RADEON R7 350 640SP SPEC
BRAND
AMD
VRAM
2 GB GDDR5
BANDWIDTH
72 GB/s
FP16 COMPUTE
1.2 TFLOPS
FP32 COMPUTE
1.2 TFLOPS
STREAM PROCESSORS
640
TDP
55 W
ARCHITECTURE
GCN 1.0
▸ AI CAPABILITY
42/ 331 models @ Q4

With 2 GB VRAM and 72 GB/s bandwidth, this GPU handles models up to 2B parameters.

Speed ≈ bandwidth / model_size × efficiency. A 7B model at Q4 runs at ~8 tok/s.

§ 01TOP MODELS FOR RADEON R7 350 640SP
42 FIT · SHOWING 20
MODELSIZEVRAM Q4TOK/SAVG
Gemma 1 2B2B1.7 GB3220.2
Granite 3.0 2B2B1.7 GB3235.8
Granite 3.1 2B2B1.7 GB3237.8
Granite 3.3 2B2B1.7 GB3220.5
Qwen 3.5 2B2B1.7 GB3229.7
Moondream2 1.9B1.9B1.6 GB3426.2
Qwen 1.5 1.8B1.8B1.6 GB3619.6
SmolLM2 1.7B1.71B1.5 GB3716.2
Qwen3 1.7B1.7B1.5 GB3830.1
stablelm-2-1_6b1.6B1.5 GB409.5
Falcon-H1 1.5B1.5B1.4 GB4343.8
GPT-2 XL 1.5B1.5B1.4 GB435.1
Qwen2.5-Coder-1.5B1.5B1.4 GB4319.6
Qwen2 Math 1.5B1.5B1.4 GB4319.6
Qwen 2.5 1.5B1.5B1.4 GB4330.2
Yi Coder 1.5B1.5B1.4 GB4314.6
DeepSeek Coder 1.3B1.3B1.3 GB4916.8
EXAONE-4.0-1.2B1.3B1.3 GB4918.9
OPT 1.3B1.3B1.3 GB495.3
Phi-1 1.3B1.3B1.3 GB497.2