AI Compatibility Tier List
Mac Studio M1 Ultra 64GB
380 AI models ranked by how well they run on your 64 GB unified memory hardware. Best match: Qwen 3.6 35B A3B (score: 98).
380
Models tested
323
Can run
64 GB
unified memory
56
Best tok/s
Tier Distribution
S 23
A 61
B 46
C 193
F 57
S: 23A: 61B: 46C: 193F: 57
S
S Tier -- Perfect Fit (23 models)
These models run flawlessly with headroom to spare.
Qwen 3.6 35B A3B35B33.3 GB56 tok/s66K ctx
moe
Qwen3-Coder 30B A3B Instruct30.5B27.9 GB67 tok/s215K ctx
moe
Qwen 3.5 35B A3B35B30.6 GB61 tok/s131K ctx
moe
Qwen3-VL 30B A3B Instruct30B27.6 GB69 tok/s218K ctx
moe
Qwen 3 30B A3B30.5B27.9 GB67 tok/s131K ctx
moe
Qwen 3.5 27B27B27.5 GB29 tok/s110K ctx
dense
Qwen 3 32B32B31.2 GB25 tok/s77K ctx
dense
Nemotron Cascade 2 30B A3B30B29.0 GB68 tok/s109K ctx
moe
Magistral Small 250724B24.9 GB32 tok/s131K ctx
dense
Qwen 3.6 27B27B25.3 GB22 tok/s262K ctx
dense
Devstral Small 2 24B Instruct24B24.9 GB32 tok/s155K ctx
dense
Nemotron 3 Nano 30B30B28.6 GB26 tok/s131K ctx
dense
GPT-OSS 20B21B23.1 GB85 tok/s128K ctx
moe
Qwen 3.5 9B9B15.5 GB86 tok/s131K ctx
dense
Qwen 3 14B14B18.8 GB56 tok/s131K ctx
dense
Devstral Small 1.124B24.9 GB32 tok/s131K ctx
dense
Phi-4-reasoning-plus 14B14.7B19.8 GB53 tok/s33K ctx
dense
Qwen 3 8B8B14.9 GB97 tok/s131K ctx
dense
Gemma 4 26B A4B25.2B26.8 GB72 tok/s100K ctx
moe
EXAONE 4.0 32B32B31.2 GB24 tok/s77K ctx
dense
Qwen 3.5 4B4B12.4 GB56 tok/s131K ctx
dense
Qwen 2.5 32B32B31.2 GB24 tok/s77K ctx
dense
Gemma 4 31B30.7B41.2 GB19 tok/s21K ctx
dense
A
A Tier -- Great Fit (61 models)
Excellent performance with comfortable memory margins.
Ornith 1.0 35B A3B
A8535.1B29.5 GB61 tok/s262K ctx
moe
Gemma 3 27B27B35.5 GB21 tok/s31K ctx
dense
OLMo 2 32B32B31.2 GB24 tok/s4K ctx
dense
Granite 4.1 30B30B30.0 GB26 tok/s82K ctx
dense
Mistral Small 3.2 24B24B24.9 GB32 tok/s131K ctx
vision
Ministral 3 14B14B18.8 GB55 tok/s195K ctx
multimodal
Codestral 2 25.0822B23.7 GB33 tok/s163K ctx
dense
Nemotron Nano 8B8B14.6 GB97 tok/s131K ctx
dense
LFM2 24B24B24.9 GB32 tok/s131K ctx
dense
CogVLM2 19B19B21.8 GB41 tok/s8K ctx
dense
Phi-4 Mini Reasoning 4B3.8B11.6 GB53 tok/s131K ctx
dense
DiffusionGemma 26B A4B25.8B27.2 GB69 tok/s98K ctx
moe
Mistral Small 24B24B24.9 GB32 tok/s33K ctx
dense
Mistral Small 3.1 24B24B24.9 GB32 tok/s131K ctx
dense
InternVL2 8B8B14.6 GB97 tok/s8K ctx
dense
Gemma 4 12B12B21.0 GB48 tok/s85K ctx
dense
Phi-4 14B14B19.4 GB55 tok/s16K ctx
dense
Kimi Linear 48B A3B48B38.0 GB15 tok/s155K ctx
linear
DeepSeek Coder V2 16B16B20.9 GB107 tok/s131K ctx
moe
Qwen 2.5 14B14B19.3 GB56 tok/s131K ctx
dense
Qwen 2.5 Coder 32B32B31.2 GB24 tok/s77K ctx
dense
Granite Code 20B20B23.2 GB39 tok/s8K ctx
dense
StarCoder 15B15B33.3 GB42 tok/s8K ctx
dense
Ministral 3 8B8B14.9 GB97 tok/s243K ctx
multimodal
Granite Code 34B34B32.2 GB23 tok/s8K ctx
dense
MiniCPM-V 2.6 8B8B14.6 GB97 tok/s2K ctx
dense
Gemma 3 12B12B20.0 GB48 tok/s101K ctx
dense
Nemotron Nano 9B v29B15.7 GB86 tok/s131K ctx
dense
CodeLlama 13B Instruct13B27.9 GB56 tok/s16K ctx
dense
Command R 35B35B31.6 GB22 tok/s111K ctx
dense
SQLCoder 7B7B14.0 GB98 tok/s8K ctx
dense
Qwen 2.5 VL 7B7B12.9 GB98 tok/s33K ctx
dense
Qwen 3 4B4B12.4 GB56 tok/s33K ctx
dense
DeepSeek R1 Distill 32B32B31.2 GB24 tok/s33K ctx
dense
LLaVA 1.6 13B13B27.9 GB56 tok/s4K ctx
dense
Magistral 7B7B14.0 GB98 tok/s8K ctx
dense
Jina Embeddings v30.57B10.9 GB8 tok/s8K ctx
dense
CodeGeeX 4 9B9B13.9 GB88 tok/s131K ctx
dense
OLMo 2 13B13B18.2 GB60 tok/s33K ctx
dense
Gemma 4 E4B8B14.0 GB74 tok/s128K ctx
dense
BGE M30.57B10.1 GB8 tok/s8K ctx
dense
CodeLlama 7B Instruct7B19.9 GB98 tok/s16K ctx
dense
StarCoder 7B7B19.4 GB98 tok/s8K ctx
dense
Nous Dolphin 13B13B29.4 GB48 tok/s16K ctx
dense
Nous Hermes 1.09B25.5 GB80 tok/s16K ctx
dense
WizardLM 13B13B27.9 GB56 tok/s8K ctx
dense
Vicuna 13B13B27.9 GB56 tok/s4K ctx
dense
Granite Code 8B8B14.6 GB97 tok/s8K ctx
dense
Qwen 2.5 7B7B12.9 GB98 tok/s131K ctx
dense
DeepSeek R1 Distill 14B14B19.3 GB56 tok/s33K ctx
dense
DevStral 7B7B14.0 GB98 tok/s8K ctx
dense
Granite 4.1 8B8B15.1 GB97 tok/s131K ctx
dense
Codestral Mamba 7B7B12.6 GB98 tok/s262K ctx
state-space
Pixtral 12B12B17.6 GB65 tok/s131K ctx
dense
LFM2.5 8B A1B8.5B13.2 GB207 tok/s128K ctx
moe
Gemma 2 27B27B35.5 GB21 tok/s8K ctx
dense
InternLM 7B7B19.9 GB98 tok/s8K ctx
dense
InternLM Chat 7B7B19.9 GB98 tok/s8K ctx
dense
mxbai Embed Large0.34B10.0 GB5 tok/s1K ctx
dense
Snowflake Arctic Embed L0.34B10.0 GB5 tok/s1K ctx
dense
Qwen 2.5 Coder 3B3B11.8 GB42 tok/s131K ctx
dense
B
B Tier -- Good Fit (46 models)
Solid performance, may need quantization or reduced context.
Llama 3.1 8B8B14.6 GB97 tok/s128K ctx
dense
GLM-4 9B9B13.9 GB88 tok/s128K ctx
dense
Baichuan 13B13B29.4 GB48 tok/s8K ctx
dense
Solar 7B7B15.0 GB98 tok/s8K ctx
dense
Gemma 4 E2B5.1B11.5 GB71 tok/s128K ctx
dense
Falcon 40B Instruct40B38.4 GB17 tok/s8K ctx
dense
LLaVA 1.5 7B7B19.9 GB98 tok/s4K ctx
dense
WizardMath 7B7B14.0 GB98 tok/s4K ctx
dense
OLMo 2 7B7B14.0 GB98 tok/s4K ctx
dense
BGE Large EN v1.50.34B10.0 GB5 tok/s1K ctx
dense
Cerebras-GPT 13B13B26.9 GB48 tok/s47K ctx
dense
Qwen 2.5 Coder 7B7B12.9 GB98 tok/s131K ctx
dense
Gemma 3 4B4B12.3 GB56 tok/s128K ctx
dense
MPT-7B-Instruct7B19.9 GB98 tok/s8K ctx
dense
Baichuan 7B7B19.9 GB98 tok/s8K ctx
dense
Nomic Embed Text v1.50.14B8.6 GB2 tok/s8K ctx
dense
Ministral 3 3B3B10.4 GB42 tok/s262K ctx
multimodal
Phi 4 Mini 4B4B11.7 GB56 tok/s128K ctx
dense
Mixtral 8x7B47B38.4 GB32 tok/s33K ctx
moe
Qwen 3.5 2B2B10.7 GB28 tok/s131K ctx
dense
DeepSeek R1 Distill 8B8B14.6 GB97 tok/s33K ctx
dense
Yi 1.5 34B34B32.2 GB23 tok/s4K ctx
dense
DeepSeek R1 Distill 7B7B12.9 GB98 tok/s33K ctx
dense
Phi 3 Mini 3.8B3.8B16.0 GB53 tok/s98K ctx
dense
Samantha 7B7B14.0 GB98 tok/s4K ctx
dense
Gemma 2 9B9B18.4 GB64 tok/s8K ctx
dense
Falcon 7B Instruct7B12.2 GB98 tok/s8K ctx
dense
Qwen 2.5 3B3B11.8 GB42 tok/s131K ctx
dense
Qwen 2.5 Coder 14B14B19.3 GB56 tok/s131K ctx
dense
Phi 3.5 Mini 4B4B16.1 GB56 tok/s98K ctx
dense
Llama 3.2 11B Vision11B16.5 GB71 tok/s16K ctx
vision
Qwen 3 1.7B1.7B10.6 GB24 tok/s33K ctx
dense
Granite 4.1 3B3B10.9 GB42 tok/s131K ctx
dense
Mistral Nemo 12B12B17.6 GB65 tok/s128K ctx
dense
All MiniLM L6 v20.02B8.1 GB2 tok/s0K ctx
dense
Granite Code 3B3B12.1 GB42 tok/s8K ctx
dense
Yi Coder 9B9B14.8 GB87 tok/s131K ctx
dense
Mistral 7B Instruct v0.37B14.0 GB98 tok/s8K ctx
dense
Phi 3 Medium 14B14B19.4 GB55 tok/s128K ctx
dense
Qwen 2.5 Coder 1.5B1.5B9.2 GB21 tok/s33K ctx
dense
Codestral 22B22B23.7 GB35 tok/s33K ctx
dense
MPT-30B-Instruct30B52.8 GB17 tok/s8K ctx
dense
InternLM 20B20B40.5 GB31 tok/s8K ctx
dense
Ministral 8B8B14.9 GB97 tok/s131K ctx
dense
Llama 3.2 3B3B11.4 GB42 tok/s128K ctx
dense
Aya Expanse 32B32B29.8 GB25 tok/s8K ctx
dense
C
C Tier -- Tight Fit (193 models)
Runs with offloading or significant compromises.
Yi 34B Chat34B32.2 GB23 tok/s77K ctx
dense
Granite 3.1 8B8B14.6 GB112 tok/s128K ctx
state-space
Yi 1.5 9B9B14.8 GB87 tok/s4K ctx
dense
StableLM 2 12B12B28.7 GB48 tok/s4K ctx
dense
SmolLM3 3B3B11.6 GB42 tok/s128K ctx
dense
Qwen3.5 35B A3B35B33.3 GB21 tok/s66K ctx
dense
Qwen3.5 35B A3B
C5235B33.3 GB21 tok/s66K ctx
dense
DeepSeek R1 1.5B1.5B9.2 GB21 tok/s33K ctx
dense
Qwen 2.5 Math 7B7B12.9 GB98 tok/s4K ctx
dense
TinyLlama 1.1B1.1B8.8 GB15 tok/s4K ctx
dense
Hermes 4.3 36B
C5236B34.0 GB20 tok/s62K ctx
dense
Vicuna 7B7B19.9 GB98 tok/s4K ctx
dense
EXAONE 4.0 32B
C5232B31.1 GB23 tok/s80K ctx
dense
aya expanse 32b heretic MPOA i1
C5232B31.1 GB23 tok/s80K ctx
dense
Baichuan M2 32B Q4 K M
C5132B31.1 GB23 tok/s80K ctx
dense
baichuan inc Baichuan M2 32B
C5132B31.1 GB23 tok/s80K ctx
dense
Qwen3.5 27B27B27.4 GB27 tok/s110K ctx
dense
gemma 3 27b it27B27.4 GB27 tok/s110K ctx
dense
gemma 3 27b it
C5127B27.4 GB27 tok/s110K ctx
dense
OpenChat 7B7B14.0 GB98 tok/s8K ctx
dense
Aya Expanse 8B8B14.6 GB97 tok/s8K ctx
dense
StarCoder2 15B15B19.8 GB45 tok/s16K ctx
dense
Mistral Small 3.2 24B Instruct 250624B25.3 GB30 tok/s134K ctx
dense
cognitivecomputations Dolphin Mistral 24B Venice Edition
C5024B25.3 GB30 tok/s134K ctx
dense
Mistral Small 24B Instruct 2501
C5024B25.3 GB30 tok/s134K ctx
dense
mistral small 3.1 24b instruct 2503 hf
C5024B25.3 GB30 tok/s134K ctx
dense
cognitivecomputations Dolphin3.0 R1 Mistral 24B
C5024B25.3 GB30 tok/s134K ctx
dense
Dolphin Mistral GLM 4.7 Flash 24B Venice Edition Thinking Uncensored i1
C5024B25.3 GB30 tok/s134K ctx
dense
cognitivecomputations Dolphin Mistral 24B Venice Edition
C5024B25.3 GB30 tok/s134K ctx
dense
Qwen 2.5 1.5B1.5B9.2 GB21 tok/s131K ctx
dense
Codestral 22B v0.1
C5022B23.8 GB33 tok/s154K ctx
dense
Codestral 22B v0.1
C5022B23.8 GB33 tok/s154K ctx
dense
Codestral 22B v0.1
C4922B23.8 GB33 tok/s154K ctx
dense
Dolphin 2.9 8B8B14.6 GB97 tok/s33K ctx
dense
DeepSeek LLM 7B7B19.4 GB98 tok/s4K ctx
dense
Codestral 22B v0.1 IMat
C4922B23.8 GB33 tok/s154K ctx
dense
OpenHermes 2.5 7B7B14.0 GB98 tok/s8K ctx
dense
Codestral 22B v0.1 i1
C4922B23.8 GB33 tok/s154K ctx
dense
Codestral 21B Pruned i1
C4921B23.1 GB34 tok/s166K ctx
dense
Zephyr 7B Beta7B14.0 GB98 tok/s33K ctx
dense
internlm2 5 20b chat
C4920B22.4 GB36 tok/s178K ctx
dense
internlm2 limarp chat 20b
C4920B22.4 GB36 tok/s178K ctx
dense
internlm2 math plus 20b i1
C4920B22.4 GB36 tok/s178K ctx
dense
Starling LM 7B7B14.0 GB98 tok/s8K ctx
dense
Qwen3.5 9B9B14.4 GB80 tok/s497K ctx
dense
Codestral RAG 19B Pruned i1
C4919B21.6 GB38 tok/s192K ctx
dense
Meta Llama 3.1 8B Instruct
C498B13.6 GB90 tok/s570K ctx
dense
Qwen3.5 9B Uncensored HauhauCS Aggressive
C499B14.4 GB80 tok/s497K ctx
dense
DeepSeek R1 Distill Qwen 14B14B18.0 GB52 tok/s290K ctx
dense
llava llama 3 8b v1 1
C488B13.6 GB90 tok/s570K ctx
dense
DeepSeek R1 0528 Qwen3 8B8B13.6 GB90 tok/s570K ctx
dense
Phi 4 reasoning vision 15B
C4815B18.7 GB48 tok/s265K ctx
dense
Qwen3.5 9B
C489B14.4 GB80 tok/s497K ctx
dense
StarCoder2 15B
C4815B18.7 GB48 tok/s265K ctx
dense
Meta Llama 3 8B Instruct
C488B13.6 GB90 tok/s570K ctx
dense
DeepSeek R1 Distill Llama 8B8B13.6 GB90 tok/s570K ctx
dense
Dolphin3.0 Llama3.1 8B
C488B13.6 GB90 tok/s570K ctx
dense
Llama 3 8B Instruct 32k v0.1
C488B13.6 GB90 tok/s570K ctx
dense
Meta Llama 3.1 8B Instruct
C488B13.6 GB90 tok/s570K ctx
dense
vntl llama3 8b v2
C488B13.6 GB90 tok/s570K ctx
dense
DeepSeek R1 0528 Qwen3 8B
C488B13.6 GB90 tok/s570K ctx
dense
gemma 3 12b it
C4812B16.5 GB60 tok/s352K ctx
dense
NousResearch Hermes 4 14B
C4814B18.0 GB52 tok/s290K ctx
dense
DeepSeek R1 0528 Qwen3 8B
C488B13.6 GB90 tok/s570K ctx
dense
Yi Coder 9B Chat
C489B14.4 GB80 tok/s497K ctx
dense
Qwen3 8B DeepSeek v3.2 Speciale Distill
C488B13.6 GB90 tok/s570K ctx
dense
LFM2.5 350M0.35B8.2 GB5 tok/s128K ctx
dense
glm 4 9b chat 1m
C489B14.4 GB80 tok/s497K ctx
dense
Llama 2 7B Chat
C487B12.9 GB98 tok/s663K ctx
dense
Hermes 3 Llama 3.1 8B
C488B13.6 GB90 tok/s570K ctx
dense
starcoder2 15b instruct v0.1
C4815B18.7 GB48 tok/s265K ctx
dense
starcoder2 15b i1
C4815B18.7 GB48 tok/s265K ctx
dense
Mistral 7B Instruct v0.2
C487B12.9 GB98 tok/s663K ctx
dense
starcoder2 15b instruct v0.1
C4815B18.7 GB48 tok/s265K ctx
dense
Hermes 2 Pro Llama 3 8B
C488B13.6 GB90 tok/s570K ctx
dense
Dolphin3.0 Llama3.1 8B
C488B13.6 GB90 tok/s570K ctx
dense
Mistral 7B Instruct v0.3
C487B12.9 GB98 tok/s663K ctx
dense
SOLAR 10.7B Instruct v1.0 uncensored
C4810.7B15.6 GB67 tok/s405K ctx
dense
granite 8b code instruct 4k
C488B13.6 GB90 tok/s570K ctx
dense
Nous Hermes 2 SOLAR 10.7B
C4810.7B15.6 GB67 tok/s405K ctx
dense
dolphin 2.9.4 llama3.1 8b
C488B13.6 GB90 tok/s570K ctx
dense
GGUF SOLARized GraniStral 14B 2102 YeAM HCT 32QKV
C4814B18.0 GB52 tok/s290K ctx
dense
Yi 1.5 9B Chat
C489B14.4 GB80 tok/s497K ctx
dense
HelpingAI 15B i1
C4815B18.7 GB48 tok/s265K ctx
dense
internlm JanusCoder 14B
C4814B18.0 GB52 tok/s290K ctx
dense
openchat 3.6 8b 20240522 IMat
C488B13.6 GB90 tok/s570K ctx
dense
GGUF SOLARized GraniStral 14B 1902 YeAM HCT
C4814B18.0 GB52 tok/s290K ctx
dense
aya expanse 8b
C488B13.6 GB90 tok/s570K ctx
dense
EXAONE 3.5 7.8B Instruct
C487.8B13.5 GB93 tok/s587K ctx
dense
Gemma 3 1B1B8.8 GB14 tok/s33K ctx
dense
dolphin v2 8b abliterated i1
C488B13.6 GB90 tok/s570K ctx
dense
EXAONE 3.5 7.8B Instruct
C487.8B13.5 GB93 tok/s587K ctx
dense
Qwen3 48B A4B Savant Commander Distill 12X Closed Open Heretic Uncensored
C4848B42.7 GB15 tok/s26K ctx
dense
EXAONE 3.5 7.8B Instruct i1
C487.8B13.5 GB93 tok/s587K ctx
dense
aya expanse 8b orthogonal heretic i1
C488B13.6 GB90 tok/s570K ctx
dense
Mistral 7B Instruct v0.3
C487B12.9 GB98 tok/s663K ctx
dense
Hermes 2 Pro Mistral 7B
C487B12.9 GB98 tok/s663K ctx
dense
exaone 3.0 7.8b it
C487.8B13.5 GB93 tok/s587K ctx
dense
zephyr 7B beta
C487B12.9 GB98 tok/s663K ctx
dense
Yi 9B Coder i1
C489B14.4 GB80 tok/s497K ctx
dense
Nous Hermes 2 Mistral 7B DPO
C487B12.9 GB98 tok/s663K ctx
dense
solar finalised finetuned Model 10.7B i1
C4810.7B15.6 GB67 tok/s405K ctx
dense
aya expanse 8b orthogonal heretic
C488B13.6 GB90 tok/s570K ctx
dense
Falcon H1R 7B
C487B12.9 GB98 tok/s663K ctx
dense
SOLAR 10.7B v1.0
C4810.7B15.6 GB67 tok/s405K ctx
dense
HelpingAI2 9B
C489B14.4 GB80 tok/s497K ctx
dense
falcon mamba 7b instruct Q4 K M
C487B12.9 GB98 tok/s663K ctx
dense
zephyr 7B alpha
C487B12.9 GB98 tok/s663K ctx
dense
speechless zephyr code functionary 7b
C487B12.9 GB98 tok/s663K ctx
dense
internlm2 5 1 8b chat i1
C488B13.6 GB90 tok/s570K ctx
dense
Gemma 2 2B2B10.6 GB28 tok/s8K ctx
dense
internlm3 8b instruct abliterated i1
C488B13.6 GB90 tok/s570K ctx
dense
StarCoder2 7B
C487B12.9 GB98 tok/s663K ctx
dense
HelpingAI2 9B i1
C489B14.4 GB80 tok/s497K ctx
dense
HelpingAI 9B 200k i1
C489B14.4 GB80 tok/s497K ctx
dense
japanese stablelm instruct gamma 7B
C487B12.9 GB98 tok/s663K ctx
dense
Neural Chat 7B7B14.0 GB98 tok/s8K ctx
dense
HelpingAI 9B i1
C489B14.4 GB80 tok/s497K ctx
dense
HelpingAI2.5 10B i1
C4810B15.1 GB72 tok/s439K ctx
dense
Falcon H1 7B Instruct
C487B12.9 GB98 tok/s663K ctx
dense
Falcon H1R 7B7B12.9 GB98 tok/s663K ctx
dense
Helply 10.2b chat i1
C4810.2B15.2 GB71 tok/s429K ctx
dense
Mamba Codestral 7B v0.1
C477B12.9 GB98 tok/s663K ctx
dense
stabilityai japanese stablelm base gamma 7b
C477B12.9 GB98 tok/s663K ctx
dense
baichuan2 7b chat
C477B12.9 GB98 tok/s663K ctx
dense
DiscoPOP zephyr 7b gemma
C477B12.9 GB98 tok/s663K ctx
dense
starcoder2 7b
C477B12.9 GB98 tok/s663K ctx
dense
zephyr 7b beta Mistral 7B Instruct v0.2
C477B12.9 GB98 tok/s663K ctx
dense
OpenChat 3.5 7B Qwen v2.0 i1
C477B12.9 GB98 tok/s663K ctx
dense
OpenChat 3.5 7B Starling v2.0 i1
C477B12.9 GB98 tok/s663K ctx
dense
internlm2 math plus 7b IMat
C477B12.9 GB98 tok/s663K ctx
dense
CodeNinja 1.0 OpenChat 7B i1
C477B12.9 GB98 tok/s663K ctx
dense
internlm2 5 7b chat i1
C477B12.9 GB98 tok/s663K ctx
dense
OpenSafetyLab MD Judge v0 2 internlm2 7b
C477B12.9 GB98 tok/s663K ctx
dense
MD Judge v0 2 internlm2 7b i1
C477B12.9 GB98 tok/s663K ctx
dense
Yi 1.5 6B Chat
C476B12.2 GB84 tok/s788K ctx
dense
zephyr 7b gemma sft african ultrachat 100k
C477B12.9 GB98 tok/s663K ctx
dense
jointpreferences mistral 7b sft helpful
C477B12.9 GB98 tok/s663K ctx
dense
zephyr 7b dpo full i1
C477B12.9 GB98 tok/s663K ctx
dense
blossom v3 baichuan2 7b i1
C477B12.9 GB98 tok/s663K ctx
dense
blossom v1 baichuan 7b i1
C477B12.9 GB98 tok/s663K ctx
dense
stablelm 2 zephyr 1 6b
C476B12.2 GB84 tok/s788K ctx
dense
Yi 1.5 6B6B12.4 GB84 tok/s4K ctx
dense
Yi 1.5 6B Chat
C476B12.2 GB84 tok/s788K ctx
dense
stablelm 2 1 6b chat imatrix
C476B12.2 GB84 tok/s788K ctx
dense
StarCoder2 7B7B12.6 GB98 tok/s16K ctx
dense
HelpingAI2 6B
C476B12.2 GB84 tok/s788K ctx
dense
HelpingAI2 6B i1
C476B12.2 GB84 tok/s788K ctx
dense
Nemotron Mini 4B4B12.2 GB56 tok/s4K ctx
dense
Qwen 2.5 Coder 0.5B0.5B8.3 GB7 tok/s131K ctx
dense
HelpingAI2.5 5B i1
C465B11.4 GB70 tok/s962K ctx
dense
Qwen3.5 4B4B10.7 GB56 tok/s1223K ctx
dense
gemma 3 4b it
C454B10.7 GB56 tok/s1223K ctx
dense
gemma 3 4b it
C454B10.7 GB56 tok/s1223K ctx
dense
Qwen 3 0.6B0.6B9.0 GB8 tok/s33K ctx
dense
Llama 3.2 1B1B8.9 GB14 tok/s128K ctx
dense
Llama 3.2 3B Instruct
C443B10.3 GB42 tok/s1643K ctx
dense
Qwen2.5 3B Instruct3B10.0 GB42 tok/s1658K ctx
dense
Llama 3.2 3B Instruct
C443B10.0 GB42 tok/s1658K ctx
dense
Ministral 3 3B Instruct 2512
C443B10.0 GB42 tok/s1658K ctx
dense
HELVETE 3B
C443B10.0 GB42 tok/s1658K ctx
dense
Hermes 3 Llama 3.2 3B
C443B10.0 GB42 tok/s1658K ctx
dense
stablelm zephyr 3b
C443B10.0 GB42 tok/s1658K ctx
dense
StarCoder2 3B
C443B10.0 GB42 tok/s1658K ctx
dense
AI21 Jamba Reasoning 3B
C443B10.0 GB42 tok/s1658K ctx
dense
stablelm 3b 4e1t
C443B10.0 GB42 tok/s1658K ctx
dense
ai21labs AI21 Jamba Reasoning 3B
C443B10.0 GB42 tok/s1658K ctx
dense
ai21labs AI21 Jamba2 3B
C443B10.0 GB42 tok/s1658K ctx
dense
HelpingAI 3B hindi i1
C433B10.0 GB42 tok/s1658K ctx
dense
AI21 Jamba2 3B
C433B10.0 GB42 tok/s1658K ctx
dense
HelpingAI 3B hindi
C433B10.0 GB42 tok/s1658K ctx
dense
AI21 Jamba2 3B i1
C433B10.0 GB42 tok/s1658K ctx
dense
gemma 2 2b it
C432B9.7 GB28 tok/s2500K ctx
dense
EXAONE 3.5 2.4B Instruct
C432.4B9.6 GB34 tok/s2094K ctx
dense
gemma 2b2B9.3 GB28 tok/s2529K ctx
dense
Gemmasutra Mini 2B v1
C432B9.3 GB28 tok/s2529K ctx
dense
StarCoder2 3B3B10.1 GB42 tok/s16K ctx
dense
gemma 2 2b it
C422B9.3 GB28 tok/s2529K ctx
dense
Qwen 3.5 0.6B0.6B9.0 GB8 tok/s131K ctx
dense
Qwen2.5 1.5B Instruct1.5B8.9 GB21 tok/s3400K ctx
dense
DeepSeek R1 Distill Qwen 1.5B1.5B8.9 GB21 tok/s3400K ctx
dense
Yi Coder 1.5B Chat
C411.5B8.9 GB21 tok/s3400K ctx
dense
stablelm 2 zephyr 1.6b
C411.6B9.0 GB22 tok/s3182K ctx
dense
logos16v2 stablelm2 1.6b i1
C411.6B9.0 GB22 tok/s3182K ctx
dense
Yi Coder 1.5B
C411.5B8.9 GB21 tok/s3400K ctx
dense
Falcon H1 1.5B Instruct1.5B8.9 GB21 tok/s3400K ctx
dense
TinyLlama 1.1B Chat v1.0
C401.1B8.6 GB15 tok/s4667K ctx
dense
DeepSeek LLM 67B67B54.5 GB9 tok/s4K ctx
dense
EXAONE 4.0 1.2B
C401.2B8.7 GB17 tok/s4271K ctx
dense
Qwen 2.5 0.5B0.5B8.3 GB7 tok/s131K ctx
dense
Llama 3.2 1B Instruct Q8 0
C401B8.7 GB14 tok/s5113K ctx
dense
TinyLlama 1.1B Chat v0.3
C401.1B8.6 GB15 tok/s4667K ctx
dense
TinyLlama 1.1B Chat v0.61.1B8.6 GB15 tok/s4667K ctx
dense
F
F Tier -- Can't Run (57 models)
Too heavy for this hardware.
Llama 3.2 1B Instruct
D401B8.5 GB14 tok/s5142K ctx
dense
gemma 3 1b it
D401B8.5 GB14 tok/s5142K ctx
dense
TinyLlama 1.1B Chat v1.0 imatrix
D401.1B8.6 GB15 tok/s4667K ctx
dense
Falcon3 1B Instruct abliterated
D391B8.5 GB14 tok/s5142K ctx
dense
SmolVLM 500M Instruct
D380.5B8.3 GB7 tok/s6057K ctx
dense
embeddinggemma 300M
D360.3B8.2 GB4 tok/s6084K ctx
dense
granite embedding 107m multilingual
D340.11B8.0 GB2 tok/s6112K ctx
dense
Falcon H1 Tiny 90M Instruct
D340.09B8.0 GB2 tok/s6114K ctx
dense
Llama 3.1 70B70B55.4 GB9 tok/s4K ctx
dense
Llama 3.3 70B70B55.4 GB9 tok/s4K ctx
dense
Qwen 2.5 72B72B56.6 GB8 tok/s4K ctx
dense
Qwen 3 235B A22B235B154.0 GB4 tok/s4K ctx
moe
Qwen3-Coder-Next80B58.1 GB21 tok/s4K ctx
moe
Qwen3-Coder 480B A35B Instruct480B303.5 GB3 tok/s4K ctx
moe
Qwen 3.5 122B A10B122B84.7 GB13 tok/s4K ctx
moe
Qwen 3.5 397B A17B397B252.9 GB4 tok/s4K ctx
moe
Mixtral 8x22B141B97.2 GB5 tok/s4K ctx
moe
Devstral 2 123B Instruct123B88.2 GB3 tok/s4K ctx
dense
Mistral Small 4 119B119B85.8 GB14 tok/s4K ctx
moe
Leanstral 119B A6B119B89.2 GB12 tok/s4K ctx
moe
Mistral Large 3675B424.9 GB2 tok/s4K ctx
moe
GLM-5744B480.7 GB2 tok/s4K ctx
moe
GLM-5.1754B486.8 GB2 tok/s4K ctx
moe
Kimi K2.51000B625.3 GB2 tok/s4K ctx
moe
Kimi K2.61000B625.3 GB2 tok/s4K ctx
moe
DeepSeek V3 671B671B476.7 GB2 tok/s4K ctx
moe
DeepSeek V3.1 671B671B476.7 GB2 tok/s4K ctx
moe
DeepSeek V4 Pro1600B871.7 GB2 tok/s4K ctx
moe
DeepSeek V4 Flash284B167.1 GB7 tok/s4K ctx
moe
DeepSeek R1 671B671B476.7 GB2 tok/s4K ctx
moe
DeepSeek Coder V2 236B236B210.4 GB6 tok/s4K ctx
moe
DeepSeek V2.5 236B236B210.4 GB6 tok/s4K ctx
moe
Llama 4 Scout 17B 16E109B77.2 GB9 tok/s4K ctx
moe
Llama 4 Maverick 17B 128E400B254.7 GB4 tok/s4K ctx
moe
Command R+ 104B104B74.7 GB4 tok/s4K ctx
dense
Command A 111B111B79.4 GB4 tok/s4K ctx
dense
Pixtral Large 124B124B88.8 GB3 tok/s4K ctx
dense
Qwen 2.5 VL 72B72B56.6 GB8 tok/s4K ctx
dense
Qwen 2.5 Math 72B72B56.6 GB8 tok/s4K ctx
dense
Nemotron 70B70B55.4 GB9 tok/s4K ctx
dense
Llama 3.1 405B405B262.6 GB2 tok/s4K ctx
dense
DeepSeek R1 Distill 70B70B55.4 GB9 tok/s4K ctx
dense
GPT-OSS 120B117B84.1 GB3 tok/s4K ctx
dense
DeepSeek V3.2671B417.6 GB2 tok/s4K ctx
moe
MiniMax M2.7
F0230B151.9 GB7 tok/s4K ctx
moe
GLM-5.2753.3B486.4 GB2 tok/s4K ctx
moe
Solar Open 100B
F0100B80.5 GB4 tok/s4K ctx
dense
K EXAONE 236B A23B
F0236B179.4 GB2 tok/s4K ctx
dense
Llama 3.3 70B Instruct
F070B58.7 GB7 tok/s4K ctx
dense
Baichuan M3 235B
F0235B178.7 GB2 tok/s4K ctx
dense
Baichuan M3 235B i1
F0235B178.7 GB2 tok/s4K ctx
dense
BaichuanMed OCR 72B i1
F072B60.2 GB7 tok/s4K ctx
dense
Solar Open 100B i1
F0100B80.5 GB4 tok/s4K ctx
dense
Solar Open 69B REAP i1
F069B58.0 GB8 tok/s4K ctx
dense
stabilityai japanese stablelm instruct beta 70b
F070B58.7 GB7 tok/s4K ctx
dense
Qwen3.5 122B A10B122B81.9 GB3 tok/s4K ctx
dense
Qwen3.5 397B A17B397B296.5 GB2 tok/s4K ctx
dense
Compare with...
See How Other Hardware Stacks Up
Check tier lists for similar GPUs and Apple Silicon configs