Skip to content

Intel Arc A770

Intel · 16GBGDDR6 · Can run 81 models

BuyAmazon
ManufacturerIntel
VRAM16 GB
Memory TypeGDDR6
ArchitectureAlchemist
Bandwidth560 GB/s
TDP225W
MSRP$349
ReleasedOct 12, 2022

AI Notes

The Intel Arc A770 offers 16GB of GDDR6 VRAM at a very competitive price, making it intriguing for local AI. It can theoretically load 13B models, though Intel GPU support in Ollama and other AI frameworks is still experimental. Early adopters should expect driver and compatibility challenges.

Compatible Models

ModelParametersBest QuantVRAM UsedFitEst. Speed
Qwen 3 0.6B600MQ4_K_M2.5 GBRuns~224 tok/s
Qwen 3.5 0.8B800MQ4_K_M1.5 GBRuns~373 tok/s
Gemma 3 1B1BQ8_02 GBRuns~280 tok/s
Llama 3.2 1B1BQ8_03 GBRuns~187 tok/s
DeepSeek R1 1.5B1.5BQ8_03 GBRuns~187 tok/s
SmolLM2 1.7B1.7BQ8_02.7 GBRuns~207 tok/s
Gemma 2 2B2BQ8_04 GBRuns~140 tok/s
Gemma 3n E2B2BQ4_K_M3.3 GBRuns~170 tok/s
Gemma 4 E2B2BQ4_K_M4 GBRuns~140 tok/s
Qwen 3.5 2B2BQ4_K_M3 GBRuns~187 tok/s
Llama 3.2 3B3BQ8_05 GBRuns~112 tok/s
StarCoder2 3B3BQ4_K_M3.5 GBRuns~160 tok/s
Phi-3 Mini 3.8B3.8BQ8_05.8 GBRuns~97 tok/s
Phi-4 Mini 3.8B3.8BQ4_K_M4.5 GBRuns~124 tok/s
Gemma 3 4B4BQ4_K_M5 GBRuns~112 tok/s
Gemma 3n E4B4BQ4_K_M4.5 GBRuns~124 tok/s
Gemma 4 E4B4BQ4_K_M6 GBRuns~93 tok/s
Qwen 3 4B4BQ4_K_M4.5 GBRuns~124 tok/s
Qwen 3.5 4B4BQ4_K_M4.5 GBRuns~124 tok/s
Yi 1.5 6B6BQ4_K_M5 GBRuns~112 tok/s
Codestral Mamba 7B7BQ4_K_M6.9 GBRuns~81 tok/s
DeepSeek R1 7B7BQ8_09 GBRuns~62 tok/s
Falcon 3 7B7BQ4_K_M6.8 GBRuns~82 tok/s
InternLM 2.5 7B7BQ4_K_M5.5 GBRuns~102 tok/s
Mistral 7B7BQ8_09 GBRuns~62 tok/s
OpenChat 3.5 7B7BQ4_K_M6.9 GBRuns~81 tok/s
Qwen 2.5 7B7BQ8_09 GBRuns~62 tok/s
Qwen 2.5 Coder 7B7BQ8_09 GBRuns~62 tok/s
Qwen 2.5 VL 7B7BQ4_K_M7 GBRuns~80 tok/s
StarCoder2 7B7BQ4_K_M5.5 GBRuns~102 tok/s
WizardLM 2 7B7BQ4_K_M6.9 GBRuns~81 tok/s
Aya Expanse 8B8BQ4_K_M6.5 GBRuns~86 tok/s
Cogito 8B8BQ4_K_M7.5 GBRuns~75 tok/s
DeepSeek R1 8B8BQ4_K_M7.5 GBRuns~75 tok/s
Dolphin 3 8B8BQ4_K_M6 GBRuns~93 tok/s
Granite 3.3 8B8BQ8_010 GBRuns~56 tok/s
Llama 3.1 8B8BQ8_010 GBRuns~56 tok/s
Nemotron 3 Nano 8B8BQ4_K_M7.5 GBRuns~75 tok/s
Nous Hermes 2 8B8BQ4_K_M6 GBRuns~93 tok/s
Qwen 3 8B8BQ4_K_M7.5 GBRuns~75 tok/s
Gemma 2 9B9BQ8_011 GBRuns~51 tok/s
Qwen 3.5 9B9BQ4_K_M7.5 GBRuns~75 tok/s
Yi 1.5 9B9BQ4_K_M6.5 GBRuns~86 tok/s
Yi Coder 9B9BQ4_K_M8 GBRuns~70 tok/s
Falcon 3 10B10BQ4_K_M8.5 GBRuns~66 tok/s
Llama 3.2 Vision 11B11BQ4_K_M8.5 GBRuns~66 tok/s
Gemma 3 12B12BQ4_K_M10.5 GBRuns~53 tok/s
Mistral Nemo 12B12BQ4_K_M9.5 GBRuns~59 tok/s
DeepSeek R1 14B14BQ4_K_M9.9 GBRuns~57 tok/s
Phi-4 14B14BQ4_K_M9.9 GBRuns~57 tok/s
Phi-4 Reasoning 14B14BQ4_K_M11 GBRuns~51 tok/s
Qwen 2.5 14B14BQ4_K_M9.9 GBRuns~57 tok/s
Qwen 2.5 Coder 14B14BQ4_K_M12 GBRuns~47 tok/s
Qwen 3 14B14BQ4_K_M12 GBRuns~47 tok/s
InternLM 2.5 20B20BQ4_K_M12 GBRuns~47 tok/s
Qwen 3.5 35B A3B35BQ4_K_M12 GBRuns~47 tok/s
gpt-oss 20B21BMXFP415 GBRuns (tight)~37 tok/s
Codestral 22B22BQ4_K_M14.7 GBRuns (tight)~38 tok/s
StarCoder2 15B15BQ8_017 GBCPU Offload~10 tok/s
Devstral 24B24BQ4_K_M17 GBCPU Offload~10 tok/s
Magistral Small 24B24BQ4_K_M17 GBCPU Offload~10 tok/s
Mistral Small 3.1 24B24BQ4_K_M18 GBCPU Offload~9 tok/s
Gemma 4 26B26BQ4_K_M20 GBCPU Offload~8 tok/s
Gemma 2 27B27BQ4_K_M17.7 GBCPU Offload~10 tok/s
Gemma 3 27B27BQ4_K_M20 GBCPU Offload~8 tok/s
Qwen 3.5 27B27BQ4_K_M19 GBCPU Offload~9 tok/s
Qwen 3.6 27B27BQ4_K_M20 GBCPU Offload~8 tok/s
Qwen 3 30B-A3B (MoE)30BQ4_K_M22 GBCPU Offload~8 tok/s
Gemma 4 31B31BQ4_K_M22 GBCPU Offload~8 tok/s
Aya Expanse 32B32BQ4_K_M22 GBCPU Offload~8 tok/s
Cogito 32B32BQ4_K_M21.5 GBCPU Offload~8 tok/s
DeepSeek R1 32B32BQ4_K_M20.7 GBCPU Offload~8 tok/s
Qwen 2.5 32B32BQ4_K_M20.7 GBCPU Offload~8 tok/s
Qwen 2.5 Coder 32B32BQ4_K_M23 GBCPU Offload~7 tok/s
Qwen 3 32B32BQ4_K_M23 GBCPU Offload~7 tok/s
QwQ 32B32BQ4_K_M21.5 GBCPU Offload~8 tok/s
Laguna XS 2.133BQ4_K_M22 GBCPU Offload~8 tok/s
WizardCoder 33B33BQ4_K_M22 GBCPU Offload~8 tok/s
Nous Hermes 2 34B34BQ4_K_M19 GBCPU Offload~9 tok/s
Yi 1.5 34B34BQ4_K_M21 GBCPU Offload~8 tok/s
Command R 35B35BQ4_K_M22.5 GBCPU Offload~8 tok/s
33 model(s) are too large for this hardware.