FitMyLLM
▸ INTEL· XE-HPG

Intel Arc A770

Running LLMs on the Arc A770 — the long read: which models fit at which quantisation, and the settings worth changing. · Or what else $349 buys

VRAM
16 GB
MID-RANGE
BANDWIDTH
512
GB/S
MODELS Q4
262/449
58%
7B Q4 SPEED
~23
GOOD
▸ 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
~55
TOK/S
7B
~23
TOK/S
14B
~12
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
512 GB/s
FP16 COMPUTE
39.3 TFLOPS
TDP
225W
MEMORY
GDDR6
ARCHITECTURE
Xe-HPG
COMPUTE UNITS
256
PCIE
Gen 4 x16
MSRP
$349
114
FAST MODELS · >30 TOK/S
Real-time chat speed
260
USABLE · >10 TOK/S
Comfortable for all tasks
262
TOTAL COMPATIBLE
Fit in VRAM at Q4
▸ WHERE TO BUY

MSRP $349

Some links are affiliate links — we may earn a small commission at no extra cost to you. This helps keep FitMyLLM free and independent.

▸ DON’T WANT TO BUY?

Test Arc A770 (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
1820
TOK/S · 3% VRAM
›
S
SmolLM2 135M0.135B
SMOLLM·2K CTX· CHAT
1214
TOK/S · 4% VRAM
›
S
nomic-embed-text-v1.5 100M0.14B
EMBEDDING·8K CTX· CHAT
1170
TOK/S · 4% VRAM
›
S
GPT-2 124M0.14B
GPT2·1K CTX· CHAT
1170
TOK/S · 4% VRAM
›
S
SmolVLM 256M0.256B
SMOLLM·8K CTX· CHAT· VISION
640
TOK/S · 4% VRAM
›
S
Gemma 3 270M0.27B
GEMMA·32K CTX· CHAT
607
TOK/S · 4% VRAM
›
S
Snowflake Arctic Embed M v2.00.305B
EMBEDDING·8K CTX· EMBEDDING· MULTILINGUAL
537
TOK/S · 4% VRAM
›
S
bge-large-en-v1.5 335M0.335B
EMBEDDING·1K CTX· CHAT
489
TOK/S · 4% VRAM
›
S
mxbai-embed-large-v10.335B
EMBEDDING·1K CTX· EMBEDDING
489
TOK/S · 4% VRAM
›
S
Snowflake Arctic Embed L0.335B
EMBEDDING·1K CTX· EMBEDDING
489
TOK/S · 4% VRAM
›
S
LFM2 350M0.35B
LFM·125K CTX· CHAT· TOOL_USE
468
TOK/S · 4% VRAM
›
S
SmolLM2 360M0.36B
SMOLLM·8K CTX· CHAT
455
TOK/S · 4% VRAM
›
S
GPT-2 Medium 345M0.38B
GPT2·1K CTX· CHAT
431
TOK/S · 5% VRAM
›
S
Qwen 2.5 0.5B0.5B
QWEN·32K CTX· CHAT
328
TOK/S · 5% VRAM
›
S
SmolVLM 500M0.5B
SMOLLM·8K CTX· CHAT· VISION
328
TOK/S · 5% VRAM
›
S
Falcon-H1 0.5B0.52B
FALCON·128K CTX· CHAT
315
TOK/S · 5% VRAM
›
S
BGE-M30.568B
EMBEDDING·8K CTX· EMBEDDING
288
TOK/S · 5% VRAM
›
S
Snowflake Arctic Embed L v2.00.568B
EMBEDDING·8K CTX· EMBEDDING· MULTILINGUAL
288
TOK/S · 5% VRAM
›
S
Qwen3-Embedding 0.6B0.6B
EMBEDDING·32K CTX· EMBEDDING· MULTILINGUAL
273
TOK/S · 5% VRAM
›
S
Falcon-H1R Tiny 0.6B0.6B
FALCON·32K CTX· CHAT· REASONING
273
TOK/S · 5% VRAM
›
S
Falcon Perception 0.6B0.6B
FALCON·4K CTX· VISION
273
TOK/S · 5% VRAM
›
S
Qwen 1.5 0.5B0.62B
QWEN·32K CTX· CHAT
264
TOK/S · 5% VRAM
›
S
LFM2 700M0.74B
LFM·125K CTX· CHAT· TOOL_USE
221
TOK/S · 6% VRAM
›
S
Qwen3 0.6B0.75B
QWEN·32K CTX· CHAT· REASONING
218
TOK/S · 6% VRAM
›
S
GPT-2 Large 774M0.81B
GPT2·1K CTX· CHAT
202
TOK/S · 6% VRAM
›
S
Qwen 3.5 0.8B0.87B
QWEN·256K CTX· CHAT· CODING· MULTILINGUAL
188
TOK/S · 6% VRAM
›
S
InternVL3 1B0.94B
OTHER·32K CTX· CHAT· VISION
174
TOK/S · 7% VRAM
›
S
MiniCPM5 1B1.08B
MINICPM·128K CTX· CHAT· REASONING· MULTILINGUAL
152
TOK/S · 7% VRAM
›
S
TinyLlama 1.1B1.1B
LLAMA·2K CTX· CHAT
149
TOK/S · 7% VRAM
›
S
LFM2.5-1.2B-Thinking1.2B
LFM·122K CTX· CHAT· REASONING· TOOL_USE
137
TOK/S · 8% VRAM
›
S
Llama-3.2-1B1.2B
LLAMA·4K CTX· CHAT
137
TOK/S · 8% VRAM
›
S
LFM2 1.2B1.2B
LFM·125K CTX· CHAT· TOOL_USE· MULTILINGUAL
137
TOK/S · 8% VRAM
›
S
Zamba2 1.2B1.2B
OTHER·4K CTX· CHAT
137
TOK/S · 8% VRAM
›
S
EXAONE-4.0-1.2B1.3B
EXAONE·64K CTX· CHAT
126
TOK/S · 8% VRAM
›
S
OPT 1.3B1.3B
OPT·2K CTX· CHAT
126
TOK/S · 8% VRAM
›
S
MiniCPM-V 4.61.3B
OTHER·256K CTX· CHAT· VISION
126
TOK/S · 8% VRAM
›
S
DeepSeek Coder 1.3B1.35B
DEEPSEEK·16K CTX· CODING
121
TOK/S · 8% VRAM
›
S
Phi-1 1.3B1.42B
PHI·2K CTX· CODING
115
TOK/S · 8% VRAM
›
S
Phi-1.5 1.3B1.42B
PHI·2K CTX· CHAT· CODING
115
TOK/S · 8% VRAM
›
S
LFM2 8B A1B8.3BMoE
LFM·125K CTX· CHAT· TOOL_USE· MULTILINGUAL
109
TOK/S · 35% VRAM
›
▸ NEXT STEP

Get personalized recommendations.

See ranked models with benchmark scores, run commands, and precise speed estimates for your Arc A770.

WHAT THIS CARD IS WORTH

Arc A770 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
5.7
8B at Q4_K_M, so cards compare like for like
VRAM PER $100
4.58 GB
what memory costs on this card
THE BIGGEST IT TAKES
ERNIE 4.5 21B A3B21.95B · IQ4_XS13.9 GB55 tok/sEST
GPT-OSS 20B21B · Q4_K_S13.8 GB44 tok/sEST
Reka Flash 321B · Q4_K_S14.2 GB8 tok/sEST
Reka Flash 3.121B · Q4_K_S14.2 GB8 tok/sEST
InternLM2 20B19.8B · Q4_K_M14.3 GB8 tok/sEST
InternLM2.5 20B19.8B · Q4_K_M14.3 GB8 tok/sEST
Ling-lite 16.8B16.8B · Q5_K_M13.6 GB54 tok/sEST
DeepSeek V2 Lite 16B16B · Q5_K_M12.9 GB54 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

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 Arc A770 — 16 GB VRAM.

▸ ARC A770 SPEC
BRAND
Intel
VRAM
16 GB GDDR6
BANDWIDTH
512 GB/s
FP16 COMPUTE
39.3 TFLOPS
FP32 COMPUTE
19.7 TFLOPS
TDP
225 W
ARCHITECTURE
Xe-HPG
MSRP
$349
▸ AI CAPABILITY
262/ 449 models @ Q4

With 16 GB VRAM and 512 GB/s bandwidth, this GPU handles models up to 19.8B parameters.

Speed ≈ bandwidth / model_size × efficiency. A 7B model at Q4 runs at ~59 tok/s.

RENT IT — LIVE PRICES
Checking what the clouds are charging…
§ 01TOP MODELS FOR ARC A770
262 FIT · SHOWING 20
MODELSIZEVRAM Q4TOK/SAVG
InternLM2 20B19.8B12.6 GB845.1
InternLM2.5 20B19.8B12.6 GB850.9
Ling-lite 16.8B16.8B10.8 GB68—
DeepSeek V2 Lite 16B16B10.3 GB6838.0
StarCoder2 15B15.96B10.2 GB1026.5
DeepSeek-Coder-V2-Lite 15.7B15.7B10.1 GB6843.0
DeepSeek-VL2 Small 16B15.7B10.1 GB6843.1
StarCoder 15B15.5B10.0 GB1121.0
InternVL3 14B15.12B9.7 GB1138.1
Phi-4-reasoning-vision 15B15B9.7 GB1142.8
DeepSeek R1 Distill Qwen 14B14.8B9.5 GB1143.9
DeepCoder 14B14.8B9.5 GB1138.7
Qwen2.5-Coder-14B14.8B9.5 GB1141.3
Qwen2.5-14B14.8B9.5 GB1141.3
Qwen3 14B14.8B9.5 GB1145.7
phi-4 14B14.66B9.4 GB1133.7
Phi-4-reasoning 14B14.66B9.4 GB1133.7
Phi-4-reasoning-plus 14B14.66B9.4 GB1175.5
Ministral 3 14B14B9.0 GB1225.9
Phi-3-medium-14b14B9.0 GB1233.7