AI Compatibility Tier List
MacBook Pro M1 Max 32GB
374 AI models ranked by how well they run on your 32 GB unified memory hardware. Best match: Qwen 3 14B (score: 93).
374
Models tested
292
Can run
32 GB
unified memory
28
Best tok/s
Tier Distribution
S 16
A 54
B 38
C 184
F 82
S: 16A: 54B: 38C: 184F: 82
S
S Tier -- Perfect Fit (16 models)
These models run flawlessly with headroom to spare.
Qwen 3 14B14B15.3 GB28 tok/s66K ctx
dense
Qwen3-VL 30B A3B Instruct30B24.1 GB32 tok/s4K ctx
moe
Phi-4-reasoning-plus 14B14.7B16.4 GB26 tok/s33K ctx
dense
Qwen 3.5 9B9B12.0 GB43 tok/s96K ctx
dense
Qwen 3 8B8B11.4 GB49 tok/s101K ctx
dense
Qwen 3.5 27B27B24.0 GB13 tok/s11K ctx
dense
GPT-OSS 20B21B19.6 GB42 tok/s38K ctx
moe
Magistral Small 250724B21.4 GB16 tok/s27K ctx
dense
Devstral Small 2 24B Instruct24B21.4 GB16 tok/s27K ctx
dense
Qwen 3.6 27B27B21.8 GB11 tok/s36K ctx
dense
Qwen 3.5 4B4B9.0 GB56 tok/s118K ctx
dense
Devstral Small 1.124B21.4 GB16 tok/s27K ctx
dense
Ministral 3 14B14B15.3 GB28 tok/s66K ctx
multimodal
Gemma 4 26B A4B25.2B23.4 GB35 tok/s14K ctx
moe
Nemotron Nano 8B8B11.2 GB49 tok/s113K ctx
dense
CogVLM2 19B19B18.4 GB20 tok/s8K ctx
dense
A
A Tier -- Great Fit (54 models)
Excellent performance with comfortable memory margins.
Phi-4 Mini Reasoning 4B3.8B8.1 GB53 tok/s131K ctx
dense
Phi-4 14B14B15.9 GB28 tok/s16K ctx
dense
DeepSeek Coder V2 16B16B17.4 GB54 tok/s43K ctx
moe
Qwen 2.5 14B14B15.8 GB28 tok/s55K ctx
dense
Qwen3-Coder 30B A3B Instruct30.5B24.4 GB30 tok/s4K ctx
moe
Codestral 2 25.0822B20.2 GB17 tok/s34K ctx
dense
InternVL2 8B8B11.2 GB49 tok/s8K ctx
dense
Mistral Small 3.2 24B24B21.4 GB16 tok/s27K ctx
vision
Gemma 3 12B12B16.6 GB24 tok/s37K ctx
dense
LFM2 24B24B21.4 GB16 tok/s27K ctx
dense
Ministral 3 8B8B11.4 GB49 tok/s101K ctx
multimodal
MiniCPM-V 2.6 8B8B11.2 GB49 tok/s2K ctx
dense
Qwen 3 30B A3B30.5B24.4 GB30 tok/s4K ctx
moe
Nemotron Nano 9B v29B12.3 GB43 tok/s86K ctx
dense
Mistral Small 24B24B21.4 GB16 tok/s27K ctx
dense
Qwen 3 4B4B9.0 GB56 tok/s33K ctx
dense
SQLCoder 7B7B10.6 GB55 tok/s8K ctx
dense
Mistral Small 3.1 24B24B21.4 GB16 tok/s27K ctx
dense
OLMo 2 13B13B14.7 GB30 tok/s33K ctx
dense
Qwen 2.5 VL 7B7B9.5 GB56 tok/s33K ctx
dense
Granite Code 20B20B19.7 GB20 tok/s8K ctx
dense
CodeLlama 7B Instruct7B16.4 GB52 tok/s16K ctx
dense
StarCoder 7B7B16.0 GB52 tok/s8K ctx
dense
Jina Embeddings v30.57B7.5 GB8 tok/s8K ctx
dense
Magistral 7B7B10.6 GB55 tok/s8K ctx
dense
Nemotron Cascade 2 30B A3B30B25.6 GB29 tok/s4K ctx
moe
Nemotron 3 Nano 30B30B25.1 GB11 tok/s4K ctx
dense
DeepSeek R1 Distill 14B14B15.8 GB28 tok/s33K ctx
dense
CodeGeeX 4 9B9B10.5 GB44 tok/s131K ctx
dense
BGE M30.57B6.6 GB8 tok/s8K ctx
dense
Gemma 4 E4B8B10.5 GB37 tok/s128K ctx
dense
InternLM 7B7B16.4 GB52 tok/s8K ctx
dense
InternLM Chat 7B7B16.4 GB52 tok/s8K ctx
dense
Granite Code 8B8B11.2 GB49 tok/s8K ctx
dense
Qwen 3.5 35B A3B35B27.2 GB26 tok/s4K ctx
moe
DevStral 7B7B10.6 GB55 tok/s8K ctx
dense
Pixtral 12B12B14.1 GB32 tok/s74K ctx
dense
Qwen 2.5 7B7B9.5 GB56 tok/s131K ctx
dense
Granite 4.1 8B8B11.7 GB49 tok/s90K ctx
dense
Codestral Mamba 7B7B9.1 GB59 tok/s262K ctx
state-space
LLaVA 1.5 7B7B16.4 GB52 tok/s4K ctx
dense
Qwen 2.5 Coder 3B3B8.4 GB42 tok/s123K ctx
dense
mxbai Embed Large0.34B6.5 GB5 tok/s1K ctx
dense
Snowflake Arctic Embed L0.34B6.5 GB5 tok/s1K ctx
dense
Nous Hermes 1.09B22.1 GB40 tok/s16K ctx
dense
Solar 7B7B11.6 GB55 tok/s8K ctx
dense
Llama 3.1 8B8B11.2 GB49 tok/s113K ctx
dense
MPT-7B-Instruct7B16.4 GB52 tok/s8K ctx
dense
Baichuan 7B7B16.4 GB52 tok/s8K ctx
dense
WizardMath 7B7B10.6 GB55 tok/s4K ctx
dense
OLMo 2 7B7B10.6 GB55 tok/s4K ctx
dense
Gemma 4 E2B5.1B8.0 GB58 tok/s128K ctx
dense
GLM-4 9B9B10.5 GB44 tok/s128K ctx
dense
BGE Large EN v1.50.34B6.5 GB5 tok/s1K ctx
dense
B
B Tier -- Good Fit (38 models)
Solid performance, may need quantization or reduced context.
Gemma 3 4B4B8.9 GB56 tok/s125K ctx
dense
Granite 4.1 30B30B26.6 GB10 tok/s4K ctx
dense
Qwen 2.5 Coder 7B7B9.5 GB56 tok/s131K ctx
dense
Phi 3 Mini 3.8B3.8B12.5 GB53 tok/s45K ctx
dense
Phi 4 Mini 4B4B8.3 GB56 tok/s128K ctx
dense
Ministral 3 3B3B6.9 GB42 tok/s262K ctx
multimodal
Phi 3.5 Mini 4B4B12.7 GB56 tok/s44K ctx
dense
Nomic Embed Text v1.50.14B5.2 GB2 tok/s8K ctx
dense
Qwen 3.5 2B2B7.3 GB28 tok/s131K ctx
dense
Qwen 2.5 Coder 14B14B15.8 GB28 tok/s55K ctx
dense
Gemma 2 9B9B15.0 GB32 tok/s8K ctx
dense
DeepSeek R1 Distill 8B8B11.2 GB49 tok/s33K ctx
dense
Samantha 7B7B10.6 GB55 tok/s4K ctx
dense
DeepSeek R1 Distill 7B7B9.5 GB56 tok/s33K ctx
dense
Qwen 2.5 3B3B8.4 GB42 tok/s123K ctx
dense
Cerebras-GPT 13B13B23.5 GB23 tok/s15K ctx
dense
Llama 3.2 11B Vision11B13.0 GB35 tok/s16K ctx
vision
CodeLlama 13B Instruct13B24.5 GB25 tok/s14K ctx
dense
Falcon 7B Instruct7B8.7 GB57 tok/s8K ctx
dense
Mistral Nemo 12B12B14.1 GB32 tok/s74K ctx
dense
Qwen 3 1.7B1.7B7.1 GB24 tok/s33K ctx
dense
Phi 3 Medium 14B14B15.9 GB28 tok/s53K ctx
dense
Granite Code 3B3B8.7 GB42 tok/s8K ctx
dense
Granite 4.1 3B3B7.4 GB42 tok/s131K ctx
dense
LLaVA 1.6 13B13B24.5 GB25 tok/s4K ctx
dense
Mistral 7B Instruct v0.37B10.6 GB55 tok/s8K ctx
dense
Yi Coder 9B9B11.3 GB44 tok/s131K ctx
dense
All MiniLM L6 v20.02B4.7 GB2 tok/s0K ctx
dense
Qwen 2.5 Coder 1.5B1.5B5.7 GB21 tok/s33K ctx
dense
WizardLM 13B13B24.5 GB25 tok/s8K ctx
dense
Vicuna 13B13B24.5 GB25 tok/s4K ctx
dense
Ministral 8B8B11.4 GB49 tok/s101K ctx
dense
Nous Dolphin 13B13B25.9 GB20 tok/s12K ctx
dense
Llama 3.2 3B3B7.9 GB42 tok/s128K ctx
dense
Codestral 22B22B20.2 GB18 tok/s33K ctx
dense
Vicuna 7B7B16.4 GB52 tok/s4K ctx
dense
Granite 3.1 8B8B11.2 GB56 tok/s113K ctx
state-space
Baichuan 13B13B25.9 GB20 tok/s8K ctx
dense
C
C Tier -- Tight Fit (184 models)
Runs with offloading or significant compromises.
SmolLM3 3B3B8.1 GB42 tok/s128K ctx
dense
Yi 1.5 9B9B11.3 GB44 tok/s4K ctx
dense
StarCoder2 15B15B16.4 GB23 tok/s16K ctx
dense
Qwen 2.5 Math 7B7B9.5 GB56 tok/s4K ctx
dense
DeepSeek LLM 7B7B16.0 GB52 tok/s4K ctx
dense
DeepSeek R1 1.5B1.5B5.7 GB21 tok/s33K ctx
dense
TinyLlama 1.1B1.1B5.4 GB15 tok/s4K ctx
dense
OpenChat 7B7B10.6 GB55 tok/s8K ctx
dense
Aya Expanse 8B8B11.2 GB49 tok/s8K ctx
dense
Phi 4 reasoning vision 15B
C5215B15.3 GB24 tok/s87K ctx
dense
StarCoder2 15B
C5215B15.3 GB24 tok/s87K ctx
dense
starcoder2 15b instruct v0.1
C5115B15.3 GB24 tok/s87K ctx
dense
starcoder2 15b i1
C5115B15.3 GB24 tok/s87K ctx
dense
starcoder2 15b instruct v0.1
C5115B15.3 GB24 tok/s87K ctx
dense
DeepSeek R1 Distill Qwen 14B14B14.5 GB26 tok/s99K ctx
dense
HelpingAI 15B i1
C5115B15.3 GB24 tok/s87K ctx
dense
Codestral RAG 19B Pruned i1
C5119B18.2 GB19 tok/s51K ctx
dense
NousResearch Hermes 4 14B
C5114B14.5 GB26 tok/s99K ctx
dense
GGUF SOLARized GraniStral 14B 2102 YeAM HCT 32QKV
C5114B14.5 GB26 tok/s99K ctx
dense
OpenHermes 2.5 7B7B10.6 GB55 tok/s8K ctx
dense
internlm JanusCoder 14B
C5114B14.5 GB26 tok/s99K ctx
dense
Dolphin 2.9 8B8B11.2 GB49 tok/s33K ctx
dense
GGUF SOLARized GraniStral 14B 1902 YeAM HCT
C5114B14.5 GB26 tok/s99K ctx
dense
Zephyr 7B Beta7B10.6 GB55 tok/s33K ctx
dense
Qwen 2.5 1.5B1.5B5.7 GB21 tok/s131K ctx
dense
gemma 3 12b it
C5112B13.1 GB30 tok/s129K ctx
dense
Starling LM 7B7B10.6 GB55 tok/s8K ctx
dense
SOLAR 10.7B Instruct v1.0 uncensored
C5010.7B12.1 GB34 tok/s155K ctx
dense
Nous Hermes 2 SOLAR 10.7B
C5010.7B12.1 GB34 tok/s155K ctx
dense
Qwen3.5 9B9B10.9 GB40 tok/s200K ctx
dense
Qwen3.5 9B Uncensored HauhauCS Aggressive
C509B10.9 GB40 tok/s200K ctx
dense
Qwen3.5 9B
C509B10.9 GB40 tok/s200K ctx
dense
solar finalised finetuned Model 10.7B i1
C5010.7B12.1 GB34 tok/s155K ctx
dense
SOLAR 10.7B v1.0
C5010.7B12.1 GB34 tok/s155K ctx
dense
Gemma 2 2B2B7.2 GB28 tok/s8K ctx
dense
Meta Llama 3.1 8B Instruct
C498B10.2 GB45 tok/s236K ctx
dense
Yi Coder 9B Chat
C499B10.9 GB40 tok/s200K ctx
dense
glm 4 9b chat 1m
C499B10.9 GB40 tok/s200K ctx
dense
llava llama 3 8b v1 1
C498B10.2 GB45 tok/s236K ctx
dense
Neural Chat 7B7B10.6 GB55 tok/s8K ctx
dense
DeepSeek R1 0528 Qwen3 8B8B10.2 GB45 tok/s236K ctx
dense
Meta Llama 3 8B Instruct
C498B10.2 GB45 tok/s236K ctx
dense
DeepSeek R1 Distill Llama 8B8B10.2 GB45 tok/s236K ctx
dense
Dolphin3.0 Llama3.1 8B
C498B10.2 GB45 tok/s236K ctx
dense
Llama 3 8B Instruct 32k v0.1
C498B10.2 GB45 tok/s236K ctx
dense
Yi 1.5 9B Chat
C499B10.9 GB40 tok/s200K ctx
dense
Helply 10.2b chat i1
C4910.2B11.8 GB35 tok/s167K ctx
dense
Meta Llama 3.1 8B Instruct
C498B10.2 GB45 tok/s236K ctx
dense
vntl llama3 8b v2
C498B10.2 GB45 tok/s236K ctx
dense
DeepSeek R1 0528 Qwen3 8B
C498B10.2 GB45 tok/s236K ctx
dense
Nemotron Mini 4B4B8.7 GB56 tok/s4K ctx
dense
DeepSeek R1 0528 Qwen3 8B
C498B10.2 GB45 tok/s236K ctx
dense
HelpingAI2.5 10B i1
C4910B11.6 GB36 tok/s172K ctx
dense
Qwen3 8B DeepSeek v3.2 Speciale Distill
C498B10.2 GB45 tok/s236K ctx
dense
Hermes 3 Llama 3.1 8B
C498B10.2 GB45 tok/s236K ctx
dense
falcon mamba 7b instruct Q4 K M
C497B9.4 GB59 tok/s281K ctx
dense
Llama 2 7B Chat
C497B9.4 GB52 tok/s281K ctx
dense
Mistral 7B Instruct v0.2
C497B9.4 GB52 tok/s281K ctx
dense
Hermes 2 Pro Llama 3 8B
C498B10.2 GB45 tok/s236K ctx
dense
Dolphin3.0 Llama3.1 8B
C498B10.2 GB45 tok/s236K ctx
dense
Yi 9B Coder i1
C499B10.9 GB40 tok/s200K ctx
dense
Mistral 7B Instruct v0.3
C497B9.4 GB52 tok/s281K ctx
dense
granite 8b code instruct 4k
C498B10.2 GB45 tok/s236K ctx
dense
dolphin 2.9.4 llama3.1 8b
C498B10.2 GB45 tok/s236K ctx
dense
Yi 1.5 6B6B9.0 GB65 tok/s4K ctx
dense
HelpingAI2 9B
C499B10.9 GB40 tok/s200K ctx
dense
openchat 3.6 8b 20240522 IMat
C498B10.2 GB45 tok/s236K ctx
dense
aya expanse 8b
C498B10.2 GB45 tok/s236K ctx
dense
Gemma 3 1B1B5.4 GB14 tok/s33K ctx
dense
HelpingAI2 9B i1
C499B10.9 GB40 tok/s200K ctx
dense
Mamba Codestral 7B v0.1
C497B9.4 GB59 tok/s281K ctx
dense
HelpingAI 9B 200k i1
C499B10.9 GB40 tok/s200K ctx
dense
dolphin v2 8b abliterated i1
C498B10.2 GB45 tok/s236K ctx
dense
HelpingAI 9B i1
C499B10.9 GB40 tok/s200K ctx
dense
aya expanse 8b orthogonal heretic i1
C498B10.2 GB45 tok/s236K ctx
dense
EXAONE 3.5 7.8B Instruct
C497.8B10.0 GB46 tok/s244K ctx
dense
Mistral 7B Instruct v0.3
C497B9.4 GB52 tok/s281K ctx
dense
Hermes 2 Pro Mistral 7B
C497B9.4 GB52 tok/s281K ctx
dense
EXAONE 3.5 7.8B Instruct
C497.8B10.0 GB46 tok/s244K ctx
dense
zephyr 7B beta
C497B9.4 GB52 tok/s281K ctx
dense
EXAONE 3.5 7.8B Instruct i1
C497.8B10.0 GB46 tok/s244K ctx
dense
Nous Hermes 2 Mistral 7B DPO
C497B9.4 GB52 tok/s281K ctx
dense
Yi 1.5 6B Chat
C496B8.7 GB60 tok/s342K ctx
dense
aya expanse 8b orthogonal heretic
C498B10.2 GB45 tok/s236K ctx
dense
exaone 3.0 7.8b it
C497.8B10.0 GB46 tok/s244K ctx
dense
Falcon H1R 7B
C497B9.4 GB52 tok/s281K ctx
dense
stablelm 2 zephyr 1 6b
C496B8.7 GB60 tok/s342K ctx
dense
zephyr 7B alpha
C497B9.4 GB52 tok/s281K ctx
dense
speechless zephyr code functionary 7b
C497B9.4 GB52 tok/s281K ctx
dense
internlm2 5 1 8b chat i1
C498B10.2 GB45 tok/s236K ctx
dense
internlm3 8b instruct abliterated i1
C498B10.2 GB45 tok/s236K ctx
dense
StarCoder2 7B
C497B9.4 GB52 tok/s281K ctx
dense
japanese stablelm instruct gamma 7B
C497B9.4 GB52 tok/s281K ctx
dense
Falcon H1 7B Instruct
C497B9.4 GB52 tok/s281K ctx
dense
Falcon H1R 7B7B9.4 GB52 tok/s281K ctx
dense
Yi 1.5 6B Chat
C486B8.7 GB60 tok/s342K ctx
dense
stablelm 2 1 6b chat imatrix
C486B8.7 GB60 tok/s342K ctx
dense
stabilityai japanese stablelm base gamma 7b
C487B9.4 GB52 tok/s281K ctx
dense
baichuan2 7b chat
C487B9.4 GB52 tok/s281K ctx
dense
DiscoPOP zephyr 7b gemma
C487B9.4 GB52 tok/s281K ctx
dense
starcoder2 7b
C487B9.4 GB52 tok/s281K ctx
dense
zephyr 7b beta Mistral 7B Instruct v0.2
C487B9.4 GB52 tok/s281K ctx
dense
OpenChat 3.5 7B Qwen v2.0 i1
C487B9.4 GB52 tok/s281K ctx
dense
OpenChat 3.5 7B Starling v2.0 i1
C487B9.4 GB52 tok/s281K ctx
dense
internlm2 math plus 7b IMat
C487B9.4 GB52 tok/s281K ctx
dense
Codestral 22B v0.1
C4822B20.4 GB16 tok/s33K ctx
dense
CodeNinja 1.0 OpenChat 7B i1
C487B9.4 GB52 tok/s281K ctx
dense
internlm2 5 7b chat i1
C487B9.4 GB52 tok/s281K ctx
dense
Codestral 22B v0.1
C4822B20.4 GB16 tok/s33K ctx
dense
OpenSafetyLab MD Judge v0 2 internlm2 7b
C487B9.4 GB52 tok/s281K ctx
dense
MD Judge v0 2 internlm2 7b i1
C487B9.4 GB52 tok/s281K ctx
dense
internlm2 5 20b chat
C4820B18.9 GB18 tok/s44K ctx
dense
zephyr 7b gemma sft african ultrachat 100k
C487B9.4 GB52 tok/s281K ctx
dense
jointpreferences mistral 7b sft helpful
C487B9.4 GB52 tok/s281K ctx
dense
zephyr 7b dpo full i1
C487B9.4 GB52 tok/s281K ctx
dense
blossom v3 baichuan2 7b i1
C487B9.4 GB52 tok/s281K ctx
dense
blossom v1 baichuan 7b i1
C487B9.4 GB52 tok/s281K ctx
dense
internlm2 limarp chat 20b
C4820B18.9 GB18 tok/s44K ctx
dense
Mistral Small 3.2 24B Instruct 250624B21.8 GB15 tok/s23K ctx
dense
HelpingAI2 6B
C486B8.7 GB60 tok/s342K ctx
dense
cognitivecomputations Dolphin Mistral 24B Venice Edition
C4824B21.8 GB15 tok/s23K ctx
dense
internlm2 math plus 20b i1
C4820B18.9 GB18 tok/s44K ctx
dense
Mistral Small 24B Instruct 2501
C4824B21.8 GB15 tok/s23K ctx
dense
HelpingAI2 6B i1
C486B8.7 GB60 tok/s342K ctx
dense
Codestral 22B v0.1
C4822B20.4 GB16 tok/s33K ctx
dense
mistral small 3.1 24b instruct 2503 hf
C4824B21.8 GB15 tok/s23K ctx
dense
cognitivecomputations Dolphin3.0 R1 Mistral 24B
C4824B21.8 GB15 tok/s23K ctx
dense
Codestral 21B Pruned i1
C4821B19.6 GB17 tok/s38K ctx
dense
StarCoder2 7B7B9.1 GB56 tok/s16K ctx
dense
HelpingAI2.5 5B i1
C485B8.0 GB70 tok/s427K ctx
dense
Dolphin Mistral GLM 4.7 Flash 24B Venice Edition Thinking Uncensored i1
C4824B21.8 GB15 tok/s23K ctx
dense
Codestral 22B v0.1 IMat
C4822B20.4 GB16 tok/s33K ctx
dense
Qwen3.5 27B27B24.0 GB12 tok/s11K ctx
dense
Codestral 22B v0.1 i1
C4822B20.4 GB16 tok/s33K ctx
dense
cognitivecomputations Dolphin Mistral 24B Venice Edition
C4824B21.8 GB15 tok/s23K ctx
dense
gemma 3 27b it27B24.0 GB12 tok/s11K ctx
dense
gemma 3 27b it
C4827B24.0 GB12 tok/s11K ctx
dense
Qwen3.5 4B4B7.3 GB56 tok/s554K ctx
dense
gemma 3 4b it
C474B7.3 GB56 tok/s554K ctx
dense
gemma 3 4b it
C474B7.3 GB56 tok/s554K ctx
dense
Qwen 2.5 Coder 0.5B0.5B4.8 GB7 tok/s131K ctx
dense
Llama 3.2 3B Instruct
C463B6.9 GB42 tok/s752K ctx
dense
Qwen 3 0.6B0.6B5.6 GB8 tok/s33K ctx
dense
Qwen2.5 3B Instruct3B6.5 GB42 tok/s767K ctx
dense
Llama 3.2 1B1B5.5 GB14 tok/s128K ctx
dense
Llama 3.2 3B Instruct
C463B6.5 GB42 tok/s767K ctx
dense
Ministral 3 3B Instruct 2512
C453B6.5 GB42 tok/s767K ctx
dense
HELVETE 3B
C453B6.5 GB42 tok/s767K ctx
dense
Hermes 3 Llama 3.2 3B
C453B6.5 GB42 tok/s767K ctx
dense
stablelm zephyr 3b
C453B6.5 GB42 tok/s767K ctx
dense
StarCoder2 3B
C453B6.5 GB42 tok/s767K ctx
dense
AI21 Jamba Reasoning 3B
C453B6.5 GB42 tok/s767K ctx
dense
stablelm 3b 4e1t
C453B6.5 GB42 tok/s767K ctx
dense
ai21labs AI21 Jamba Reasoning 3B
C453B6.5 GB42 tok/s767K ctx
dense
ai21labs AI21 Jamba2 3B
C453B6.5 GB42 tok/s767K ctx
dense
HelpingAI 3B hindi i1
C453B6.5 GB42 tok/s767K ctx
dense
AI21 Jamba2 3B
C453B6.5 GB42 tok/s767K ctx
dense
HelpingAI 3B hindi
C453B6.5 GB42 tok/s767K ctx
dense
AI21 Jamba2 3B i1
C453B6.5 GB42 tok/s767K ctx
dense
gemma 2 2b it
C442B6.2 GB28 tok/s1164K ctx
dense
EXAONE 3.5 2.4B Instruct
C442.4B6.1 GB34 tok/s980K ctx
dense
StarCoder2 3B3B6.6 GB42 tok/s16K ctx
dense
gemma 2b2B5.8 GB28 tok/s1192K ctx
dense
Gemmasutra Mini 2B v1
C442B5.8 GB28 tok/s1192K ctx
dense
gemma 2 2b it
C442B5.8 GB28 tok/s1192K ctx
dense
Qwen 3.5 0.6B0.6B5.6 GB8 tok/s131K ctx
dense
Qwen2.5 1.5B Instruct1.5B5.4 GB21 tok/s1617K ctx
dense
DeepSeek R1 Distill Qwen 1.5B1.5B5.4 GB21 tok/s1617K ctx
dense
Yi Coder 1.5B Chat
C421.5B5.4 GB21 tok/s1617K ctx
dense
Aya Expanse 32B32B26.3 GB10 tok/s4K ctx
dense
stablelm 2 zephyr 1.6b
C421.6B5.5 GB22 tok/s1511K ctx
dense
logos16v2 stablelm2 1.6b i1
C421.6B5.5 GB22 tok/s1511K ctx
dense
Yi Coder 1.5B
C421.5B5.4 GB21 tok/s1617K ctx
dense
Falcon H1 1.5B Instruct1.5B5.4 GB21 tok/s1617K ctx
dense
TinyLlama 1.1B Chat v1.0
C411.1B5.2 GB15 tok/s2236K ctx
dense
Llama 3.2 1B Instruct Q8 0
C411B5.3 GB14 tok/s2439K ctx
dense
EXAONE 4.0 1.2B
C411.2B5.2 GB17 tok/s2043K ctx
dense
TinyLlama 1.1B Chat v0.3
C411.1B5.2 GB15 tok/s2236K ctx
dense
TinyLlama 1.1B Chat v0.61.1B5.2 GB15 tok/s2236K ctx
dense
Qwen 2.5 0.5B0.5B4.8 GB7 tok/s131K ctx
dense
Llama 3.2 1B Instruct
C411B5.1 GB14 tok/s2468K ctx
dense
gemma 3 1b it
C411B5.1 GB14 tok/s2468K ctx
dense
TinyLlama 1.1B Chat v1.0 imatrix
C411.1B5.2 GB15 tok/s2236K ctx
dense
Falcon3 1B Instruct abliterated
C401B5.1 GB14 tok/s2468K ctx
dense
F
F Tier -- Can't Run (82 models)
Too heavy for this hardware.
StableLM 2 12B12B25.2 GB20 tok/s4K ctx
dense
SmolVLM 500M Instruct
D390.5B4.9 GB7 tok/s2924K ctx
dense
embeddinggemma 300M
D370.3B4.7 GB4 tok/s2950K ctx
dense
granite embedding 107m multilingual
D350.11B4.5 GB2 tok/s2979K ctx
dense
Falcon H1 Tiny 90M Instruct
D340.09B4.5 GB2 tok/s2981K ctx
dense
EXAONE 4.0 32B
D3032B27.6 GB9 tok/s4K ctx
dense
aya expanse 32b heretic MPOA i1
D3032B27.6 GB9 tok/s4K ctx
dense
Baichuan M2 32B Q4 K M
D3032B27.6 GB9 tok/s4K ctx
dense
baichuan inc Baichuan M2 32B
D3032B27.6 GB9 tok/s4K ctx
dense
Llama 3.1 70B70B51.9 GB3 tok/s4K ctx
dense
Llama 3.3 70B70B51.9 GB3 tok/s4K ctx
dense
Qwen 2.5 32B32B27.8 GB9 tok/s4K ctx
dense
Qwen 2.5 72B72B53.2 GB3 tok/s4K ctx
dense
Qwen 2.5 Coder 32B32B27.8 GB9 tok/s4K ctx
dense
Qwen 3 32B32B27.8 GB9 tok/s4K ctx
dense
Qwen 3 235B A22B235B150.6 GB2 tok/s4K ctx
moe
Qwen3-Coder-Next80B54.6 GB7 tok/s4K ctx
moe
Qwen3-Coder 480B A35B Instruct480B300.0 GB2 tok/s4K ctx
moe
Qwen 3.6 35B A3B35B29.8 GB21 tok/s4K ctx
moe
Qwen 3.5 122B A10B122B81.2 GB6 tok/s4K ctx
moe
Qwen 3.5 397B A17B397B249.4 GB2 tok/s4K ctx
moe
Gemma 2 27B27B32.1 GB7 tok/s4K ctx
dense
Gemma 3 27B27B32.1 GB7 tok/s4K ctx
dense
Mixtral 8x7B47B35.0 GB9 tok/s4K ctx
moe
Mixtral 8x22B141B93.8 GB2 tok/s4K ctx
moe
DeepSeek R1 Distill 32B32B27.8 GB9 tok/s4K ctx
dense
Command R 35B35B28.1 GB8 tok/s4K ctx
dense
Falcon 40B Instruct40B35.0 GB5 tok/s4K ctx
dense
StarCoder 15B15B29.8 GB14 tok/s8K ctx
dense
MPT-30B-Instruct30B49.4 GB5 tok/s4K ctx
dense
Devstral 2 123B Instruct123B84.8 GB2 tok/s4K ctx
dense
Mistral Small 4 119B119B82.3 GB7 tok/s4K ctx
moe
Leanstral 119B A6B119B85.7 GB6 tok/s4K ctx
moe
Mistral Large 3675B421.5 GB2 tok/s4K ctx
moe
GLM-5744B477.2 GB2 tok/s4K ctx
moe
GLM-5.1754B483.3 GB2 tok/s4K ctx
moe
Kimi K2.51000B621.8 GB2 tok/s4K ctx
moe
Kimi K2.61000B621.8 GB2 tok/s4K ctx
moe
Kimi Linear 48B A3B48B34.6 GB4 tok/s4K ctx
linear
InternLM 20B20B37.1 GB8 tok/s4K ctx
dense
Yi 34B Chat34B28.8 GB8 tok/s4K ctx
dense
DeepSeek V3 671B671B473.2 GB2 tok/s4K ctx
moe
DeepSeek V3.1 671B671B473.2 GB2 tok/s4K ctx
moe
DeepSeek V4 Pro1600B868.2 GB2 tok/s4K ctx
moe
DeepSeek V4 Flash284B163.7 GB3 tok/s4K ctx
moe
DeepSeek R1 671B671B473.2 GB2 tok/s4K ctx
moe
DeepSeek Coder V2 236B236B206.9 GB3 tok/s4K ctx
moe
DeepSeek V2.5 236B236B206.9 GB3 tok/s4K ctx
moe
DeepSeek LLM 67B67B51.0 GB3 tok/s4K ctx
dense
Llama 4 Scout 17B 16E109B73.8 GB4 tok/s4K ctx
moe
Llama 4 Maverick 17B 128E400B251.3 GB2 tok/s4K ctx
moe
Command R+ 104B104B71.2 GB2 tok/s4K ctx
dense
Command A 111B111B76.0 GB2 tok/s4K ctx
dense
Yi 1.5 34B34B28.8 GB8 tok/s4K ctx
dense
Pixtral Large 124B124B85.4 GB2 tok/s4K ctx
dense
Qwen 2.5 VL 72B72B53.2 GB3 tok/s4K ctx
dense
Qwen 2.5 Math 72B72B53.2 GB3 tok/s4K ctx
dense
Nemotron 70B70B51.9 GB3 tok/s4K ctx
dense
OLMo 2 32B32B27.8 GB9 tok/s4K ctx
dense
EXAONE 4.0 32B32B27.8 GB9 tok/s4K ctx
dense
Llama 3.1 405B405B259.1 GB2 tok/s4K ctx
dense
DeepSeek R1 Distill 70B70B51.9 GB3 tok/s4K ctx
dense
Granite Code 34B34B28.8 GB8 tok/s4K ctx
dense
GPT-OSS 120B117B80.6 GB2 tok/s4K ctx
dense
Gemma 4 31B30.7B37.7 GB5 tok/s4K ctx
dense
DeepSeek V3.2671B414.1 GB2 tok/s4K ctx
moe
MiniMax M2.7
F0230B148.4 GB4 tok/s4K ctx
moe
Solar Open 100B
F0100B77.1 GB2 tok/s4K ctx
dense
Qwen3 48B A4B Savant Commander Distill 12X Closed Open Heretic Uncensored
F048B39.3 GB4 tok/s4K ctx
dense
K EXAONE 236B A23B
F0236B176.0 GB2 tok/s4K ctx
dense
Qwen3.5 35B A3B
F035B29.8 GB7 tok/s4K ctx
dense
Llama 3.3 70B Instruct
F070B55.3 GB2 tok/s4K ctx
dense
Baichuan M3 235B
F0235B175.2 GB2 tok/s4K ctx
dense
Baichuan M3 235B i1
F0235B175.2 GB2 tok/s4K ctx
dense
BaichuanMed OCR 72B i1
F072B56.7 GB2 tok/s4K ctx
dense
Solar Open 100B i1
F0100B77.1 GB2 tok/s4K ctx
dense
Solar Open 69B REAP i1
F069B54.5 GB2 tok/s4K ctx
dense
Hermes 4.3 36B
F036B30.5 GB7 tok/s4K ctx
dense
stabilityai japanese stablelm instruct beta 70b
F070B55.3 GB2 tok/s4K ctx
dense
Qwen3.5 122B A10B122B78.4 GB2 tok/s4K ctx
dense
Qwen3.5 35B A3B35B29.8 GB7 tok/s4K ctx
dense
Qwen3.5 397B A17B397B293.0 GB2 tok/s4K ctx
dense
Compare with...
See How Other Hardware Stacks Up
Check tier lists for similar GPUs and Apple Silicon configs