AI Compatibility Tier List
NVIDIA L40 48GB
374 AI models ranked by how well they run on your 48 GB VRAM hardware. Best match: Qwen 3.6 35B A3B (score: 98).
374
Models tested
320
Can run
48 GB
VRAM
92
Best tok/s
Tier Distribution
S 22
A 59
B 51
C 188
F 54
S: 22A: 59B: 51C: 188F: 54
S
S Tier -- Perfect Fit (22 models)
These models run flawlessly with headroom to spare.
Qwen 3.6 35B A3B35B31.2 GB92 tok/s82K ctx
moe
Qwen 3.5 35B A3B35B28.5 GB100 tok/s131K ctx
moe
Qwen3-Coder 30B A3B Instruct30.5B25.8 GB73 tok/s256K ctx
moe
Qwen3-VL 30B A3B Instruct30B25.5 GB105 tok/s256K ctx
moe
Qwen 3 30B A3B30.5B25.8 GB73 tok/s131K ctx
moe
Qwen 3.5 27B27B25.3 GB31 tok/s130K ctx
dense
Nemotron Cascade 2 30B A3B30B26.9 GB104 tok/s131K ctx
moe
Qwen 3 32B32B29.1 GB19 tok/s93K ctx
dense
Magistral Small 250724B22.8 GB35 tok/s131K ctx
dense
Devstral Small 2 24B Instruct24B22.8 GB35 tok/s181K ctx
dense
Qwen 3 14B14B16.7 GB98 tok/s131K ctx
dense
GPT-OSS 20B21B21.0 GB116 tok/s128K ctx
moe
Qwen 3.6 27B27B23.1 GB20 tok/s262K ctx
dense
Qwen 3.5 9B9B13.4 GB111 tok/s131K ctx
dense
Nemotron 3 Nano 30B30B26.4 GB23 tok/s131K ctx
dense
Phi-4-reasoning-plus 14B14.7B17.7 GB83 tok/s33K ctx
dense
Devstral Small 1.124B22.8 GB35 tok/s131K ctx
dense
Gemma 4 26B A4B25.2B24.7 GB99 tok/s118K ctx
moe
Gemma 4 31B30.7B39.1 GB15 tok/s26K ctx
dense
Qwen 3 8B8B12.8 GB120 tok/s131K ctx
dense
EXAONE 4.0 32B32B29.1 GB22 tok/s93K ctx
dense
Qwen 3.5 4B4B10.3 GB60 tok/s131K ctx
dense
A
A Tier -- Great Fit (59 models)
Excellent performance with comfortable memory margins.
Ministral 3 14B14B16.7 GB89 tok/s221K ctx
multimodal
Gemma 3 27B27B33.4 GB19 tok/s37K ctx
dense
Qwen 2.5 32B32B29.1 GB22 tok/s93K ctx
dense
Kimi Linear 48B A3B48B35.9 GB14 tok/s225K ctx
linear
Mistral Small 3.2 24B24B22.8 GB35 tok/s131K ctx
vision
Codestral 2 25.0822B21.6 GB37 tok/s189K ctx
dense
Nemotron Nano 8B8B12.5 GB128 tok/s131K ctx
dense
OLMo 2 32B32B29.1 GB22 tok/s4K ctx
dense
LFM2 24B24B22.8 GB35 tok/s131K ctx
dense
Granite 4.1 30B30B27.9 GB23 tok/s98K ctx
dense
CogVLM2 19B19B19.7 GB52 tok/s8K ctx
dense
Phi-4 Mini Reasoning 4B3.8B9.5 GB61 tok/s131K ctx
dense
Phi-4 14B14B17.3 GB89 tok/s16K ctx
dense
Mistral Small 24B24B22.8 GB35 tok/s33K ctx
dense
Qwen 2.5 14B14B17.2 GB90 tok/s131K ctx
dense
Mistral Small 3.1 24B24B22.8 GB35 tok/s131K ctx
dense
StarCoder 15B15B31.1 GB65 tok/s8K ctx
dense
InternVL2 8B8B12.5 GB128 tok/s8K ctx
dense
Qwen3-Coder-Next80B56.0 GB25 tok/s4K ctx
moe
DeepSeek Coder V2 16B16B18.8 GB148 tok/s131K ctx
moe
Granite Code 20B20B21.1 GB48 tok/s8K ctx
dense
Gemma 3 12B12B17.9 GB69 tok/s115K ctx
dense
CodeLlama 13B Instruct13B25.8 GB89 tok/s16K ctx
dense
Ministral 3 8B8B12.8 GB128 tok/s262K ctx
multimodal
MiniCPM-V 2.6 8B8B12.5 GB128 tok/s2K ctx
dense
Qwen 2.5 Coder 32B32B29.1 GB22 tok/s93K ctx
dense
Nemotron Nano 9B v29B13.6 GB121 tok/s131K ctx
dense
Granite Code 34B34B30.1 GB20 tok/s8K ctx
dense
SQLCoder 7B7B11.9 GB112 tok/s8K ctx
dense
LLaVA 1.6 13B13B25.8 GB89 tok/s4K ctx
dense
Qwen 3 4B4B10.3 GB60 tok/s33K ctx
dense
Command R 35B35B29.5 GB20 tok/s131K ctx
dense
Qwen 2.5 VL 7B7B10.8 GB112 tok/s33K ctx
dense
OLMo 2 13B13B16.1 GB96 tok/s33K ctx
dense
DeepSeek R1 Distill 32B32B29.1 GB22 tok/s33K ctx
dense
Qwen 2.5 VL 72B72B54.5 GB6 tok/s4K ctx
dense
Jina Embeddings v30.57B8.8 GB9 tok/s8K ctx
dense
Magistral 7B7B11.9 GB112 tok/s8K ctx
dense
CodeGeeX 4 9B9B11.8 GB124 tok/s131K ctx
dense
Gemma 4 E4B8B11.9 GB93 tok/s128K ctx
dense
Nous Dolphin 13B13B27.3 GB77 tok/s16K ctx
dense
BGE M30.57B8.0 GB9 tok/s8K ctx
dense
DeepSeek R1 Distill 14B14B17.2 GB90 tok/s33K ctx
dense
WizardLM 13B13B25.8 GB89 tok/s8K ctx
dense
Vicuna 13B13B25.8 GB89 tok/s4K ctx
dense
CodeLlama 7B Instruct7B17.8 GB112 tok/s16K ctx
dense
Nous Hermes 1.09B23.4 GB113 tok/s16K ctx
dense
StarCoder 7B7B17.3 GB112 tok/s8K ctx
dense
Qwen 2.5 7B7B10.8 GB112 tok/s131K ctx
dense
Granite Code 8B8B12.5 GB128 tok/s8K ctx
dense
Pixtral 12B12B15.5 GB104 tok/s131K ctx
dense
DevStral 7B7B11.9 GB112 tok/s8K ctx
dense
Granite 4.1 8B8B13.0 GB128 tok/s131K ctx
dense
Falcon 40B Instruct40B36.3 GB15 tok/s8K ctx
dense
Codestral Mamba 7B7B10.5 GB112 tok/s262K ctx
state-space
Gemma 2 27B27B33.4 GB19 tok/s8K ctx
dense
mxbai Embed Large0.34B7.9 GB5 tok/s1K ctx
dense
InternLM 7B7B17.8 GB112 tok/s8K ctx
dense
InternLM Chat 7B7B17.8 GB112 tok/s8K ctx
dense
B
B Tier -- Good Fit (51 models)
Solid performance, may need quantization or reduced context.
Mixtral 8x7B47B36.3 GB44 tok/s33K ctx
moe
Snowflake Arctic Embed L0.34B7.9 GB5 tok/s1K ctx
dense
Qwen 2.5 Coder 3B3B9.7 GB48 tok/s131K ctx
dense
Baichuan 13B13B27.3 GB77 tok/s8K ctx
dense
Llama 3.3 70B70B53.3 GB6 tok/s4K ctx
dense
Gemma 4 E2B5.1B9.3 GB82 tok/s128K ctx
dense
GLM-4 9B9B11.8 GB124 tok/s128K ctx
dense
Cerebras-GPT 13B13B24.8 GB77 tok/s54K ctx
dense
Llama 3.1 8B8B12.5 GB128 tok/s128K ctx
dense
Solar 7B7B12.9 GB112 tok/s8K ctx
dense
BGE Large EN v1.50.34B7.9 GB5 tok/s1K ctx
dense
WizardMath 7B7B11.9 GB112 tok/s4K ctx
dense
OLMo 2 7B7B11.9 GB112 tok/s4K ctx
dense
LLaVA 1.5 7B7B17.8 GB112 tok/s4K ctx
dense
Qwen 2.5 Coder 7B7B10.8 GB112 tok/s131K ctx
dense
Gemma 3 4B4B10.2 GB64 tok/s128K ctx
dense
Llama 3.1 70B70B53.3 GB6 tok/s4K ctx
dense
Qwen 2.5 72B72B54.5 GB6 tok/s4K ctx
dense
Nomic Embed Text v1.50.14B6.5 GB2 tok/s8K ctx
dense
MPT-7B-Instruct7B17.8 GB112 tok/s8K ctx
dense
Baichuan 7B7B17.8 GB112 tok/s8K ctx
dense
Ministral 3 3B3B8.3 GB48 tok/s262K ctx
multimodal
Phi 4 Mini 4B4B9.6 GB64 tok/s128K ctx
dense
Qwen 3.5 2B2B8.6 GB30 tok/s131K ctx
dense
DeepSeek R1 Distill 8B8B12.5 GB128 tok/s33K ctx
dense
Phi 3 Mini 3.8B3.8B13.9 GB61 tok/s109K ctx
dense
Qwen 2.5 Coder 14B14B17.2 GB90 tok/s131K ctx
dense
DeepSeek R1 Distill 7B7B10.8 GB112 tok/s33K ctx
dense
Samantha 7B7B11.9 GB112 tok/s4K ctx
dense
Gemma 2 9B9B16.3 GB81 tok/s8K ctx
dense
Yi 1.5 34B34B30.1 GB20 tok/s4K ctx
dense
Qwen 2.5 3B3B9.7 GB48 tok/s131K ctx
dense
Llama 3.2 11B Vision11B14.4 GB113 tok/s16K ctx
vision
Phi 3.5 Mini 4B4B14.0 GB64 tok/s109K ctx
dense
Falcon 7B Instruct7B10.1 GB112 tok/s8K ctx
dense
InternLM 20B20B38.4 GB39 tok/s8K ctx
dense
DeepSeek R1 Distill 70B70B53.3 GB6 tok/s4K ctx
dense
Qwen 3 1.7B1.7B8.4 GB26 tok/s33K ctx
dense
Mistral Nemo 12B12B15.5 GB104 tok/s128K ctx
dense
Granite 4.1 3B3B8.8 GB48 tok/s131K ctx
dense
Phi 3 Medium 14B14B17.3 GB89 tok/s128K ctx
dense
Granite Code 3B3B10.0 GB48 tok/s8K ctx
dense
All MiniLM L6 v20.02B6.0 GB2 tok/s0K ctx
dense
Yi Coder 9B9B12.7 GB123 tok/s131K ctx
dense
Qwen 2.5 Coder 1.5B1.5B7.0 GB24 tok/s33K ctx
dense
Mistral 7B Instruct v0.37B11.9 GB112 tok/s8K ctx
dense
Codestral 22B22B21.6 GB39 tok/s33K ctx
dense
MPT-30B-Instruct30B50.7 GB12 tok/s8K ctx
dense
Ministral 8B8B12.8 GB128 tok/s131K ctx
dense
Nemotron 70B70B53.3 GB6 tok/s4K ctx
dense
Llama 3.2 3B3B9.2 GB48 tok/s128K ctx
dense
C
C Tier -- Tight Fit (188 models)
Runs with offloading or significant compromises.
Aya Expanse 32B32B27.7 GB22 tok/s8K ctx
dense
StableLM 2 12B12B26.5 GB72 tok/s4K ctx
dense
Granite 3.1 8B8B12.5 GB128 tok/s128K ctx
state-space
Yi 1.5 9B9B12.7 GB123 tok/s4K ctx
dense
Yi 34B Chat34B30.1 GB20 tok/s94K ctx
dense
SmolLM3 3B3B9.5 GB48 tok/s128K ctx
dense
DeepSeek R1 1.5B1.5B7.0 GB24 tok/s33K ctx
dense
TinyLlama 1.1B1.1B6.7 GB18 tok/s4K ctx
dense
Qwen3.5 35B A3B35B31.2 GB18 tok/s82K ctx
dense
Qwen 2.5 Math 7B7B10.8 GB112 tok/s4K ctx
dense
Qwen3.5 35B A3B
C5135B31.2 GB18 tok/s82K ctx
dense
Hermes 4.3 36B
C5136B31.9 GB18 tok/s77K ctx
dense
Vicuna 7B7B17.8 GB112 tok/s4K ctx
dense
StarCoder2 15B15B17.7 GB71 tok/s16K ctx
dense
EXAONE 4.0 32B
C5032B29.0 GB20 tok/s97K ctx
dense
aya expanse 32b heretic MPOA i1
C5032B29.0 GB20 tok/s97K ctx
dense
Baichuan M2 32B Q4 K M
C5032B29.0 GB20 tok/s97K ctx
dense
baichuan inc Baichuan M2 32B
C4932B29.0 GB20 tok/s97K ctx
dense
Qwen3.5 27B27B25.3 GB26 tok/s131K ctx
dense
gemma 3 27b it27B25.3 GB25 tok/s131K ctx
dense
OpenChat 7B7B11.9 GB112 tok/s8K ctx
dense
gemma 3 27b it
C4927B25.3 GB25 tok/s131K ctx
dense
Mistral Small 3.2 24B Instruct 250624B23.2 GB33 tok/s157K ctx
dense
Aya Expanse 8B8B12.5 GB128 tok/s8K ctx
dense
cognitivecomputations Dolphin Mistral 24B Venice Edition
C4924B23.2 GB33 tok/s157K ctx
dense
Mistral Small 24B Instruct 2501
C4924B23.2 GB33 tok/s157K ctx
dense
mistral small 3.1 24b instruct 2503 hf
C4924B23.2 GB33 tok/s157K ctx
dense
cognitivecomputations Dolphin3.0 R1 Mistral 24B
C4924B23.2 GB33 tok/s157K ctx
dense
Qwen 2.5 1.5B1.5B7.0 GB24 tok/s131K ctx
dense
Dolphin Mistral GLM 4.7 Flash 24B Venice Edition Thinking Uncensored i1
C4924B23.2 GB33 tok/s157K ctx
dense
DeepSeek R1 Distill Qwen 14B14B15.9 GB83 tok/s329K ctx
dense
Codestral 22B v0.1
C4922B21.7 GB38 tok/s179K ctx
dense
Codestral 22B v0.1
C4922B21.7 GB38 tok/s179K ctx
dense
Phi 4 reasoning vision 15B
C4915B16.6 GB75 tok/s302K ctx
dense
StarCoder2 15B
C4915B16.6 GB75 tok/s302K ctx
dense
NousResearch Hermes 4 14B
C4914B15.9 GB83 tok/s329K ctx
dense
cognitivecomputations Dolphin Mistral 24B Venice Edition
C4924B23.2 GB33 tok/s157K ctx
dense
Codestral 22B v0.1
C4822B21.7 GB38 tok/s179K ctx
dense
Qwen 2.5 Math 72B72B54.5 GB6 tok/s4K ctx
dense
gemma 3 12b it
C4812B14.4 GB97 tok/s398K ctx
dense
starcoder2 15b instruct v0.1
C4815B16.6 GB75 tok/s302K ctx
dense
Codestral 22B v0.1 IMat
C4822B21.7 GB38 tok/s179K ctx
dense
GGUF SOLARized GraniStral 14B 2102 YeAM HCT 32QKV
C4814B15.9 GB83 tok/s329K ctx
dense
starcoder2 15b i1
C4815B16.6 GB75 tok/s302K ctx
dense
starcoder2 15b instruct v0.1
C4815B16.6 GB75 tok/s302K ctx
dense
internlm JanusCoder 14B
C4814B15.9 GB83 tok/s329K ctx
dense
GGUF SOLARized GraniStral 14B 1902 YeAM HCT
C4814B15.9 GB83 tok/s329K ctx
dense
HelpingAI 15B i1
C4815B16.6 GB75 tok/s302K ctx
dense
Codestral 22B v0.1 i1
C4822B21.7 GB38 tok/s179K ctx
dense
Codestral 21B Pruned i1
C4821B21.0 GB41 tok/s192K ctx
dense
internlm2 5 20b chat
C4820B20.2 GB45 tok/s205K ctx
dense
internlm2 limarp chat 20b
C4820B20.2 GB45 tok/s205K ctx
dense
Dolphin 2.9 8B8B12.5 GB128 tok/s33K ctx
dense
internlm2 math plus 20b i1
C4820B20.2 GB45 tok/s205K ctx
dense
Codestral RAG 19B Pruned i1
C4819B19.5 GB49 tok/s221K ctx
dense
OpenHermes 2.5 7B7B11.9 GB112 tok/s8K ctx
dense
DeepSeek LLM 7B7B17.3 GB112 tok/s4K ctx
dense
SOLAR 10.7B Instruct v1.0 uncensored
C4810.7B13.5 GB108 tok/s456K ctx
dense
Nous Hermes 2 SOLAR 10.7B
C4810.7B13.5 GB108 tok/s456K ctx
dense
Zephyr 7B Beta7B11.9 GB112 tok/s33K ctx
dense
Qwen3.5 9B9B12.2 GB113 tok/s558K ctx
dense
Qwen3.5 9B Uncensored HauhauCS Aggressive
C489B12.2 GB113 tok/s558K ctx
dense
solar finalised finetuned Model 10.7B i1
C4810.7B13.5 GB108 tok/s456K ctx
dense
Qwen3.5 9B
C489B12.2 GB113 tok/s558K ctx
dense
Qwen3 48B A4B Savant Commander Distill 12X Closed Open Heretic Uncensored
C4848B40.6 GB13 tok/s37K ctx
dense
SOLAR 10.7B v1.0
C4710.7B13.5 GB108 tok/s456K ctx
dense
Starling LM 7B7B11.9 GB112 tok/s8K ctx
dense
Yi Coder 9B Chat
C479B12.2 GB113 tok/s558K ctx
dense
glm 4 9b chat 1m
C479B12.2 GB113 tok/s558K ctx
dense
Meta Llama 3.1 8B Instruct
C478B11.5 GB127 tok/s639K ctx
dense
llava llama 3 8b v1 1
C478B11.5 GB127 tok/s639K ctx
dense
DeepSeek R1 0528 Qwen3 8B8B11.5 GB127 tok/s639K ctx
dense
Gemma 3 1B1B6.7 GB16 tok/s33K ctx
dense
Meta Llama 3 8B Instruct
C478B11.5 GB127 tok/s639K ctx
dense
Yi 1.5 9B Chat
C479B12.2 GB113 tok/s558K ctx
dense
DeepSeek R1 Distill Llama 8B8B11.5 GB127 tok/s639K ctx
dense
Dolphin3.0 Llama3.1 8B
C478B11.5 GB127 tok/s639K ctx
dense
Llama 3 8B Instruct 32k v0.1
C478B11.5 GB127 tok/s639K ctx
dense
Helply 10.2b chat i1
C4710.2B13.1 GB114 tok/s483K ctx
dense
Meta Llama 3.1 8B Instruct
C478B11.5 GB127 tok/s639K ctx
dense
vntl llama3 8b v2
C478B11.5 GB127 tok/s639K ctx
dense
HelpingAI2.5 10B i1
C4710B13.0 GB102 tok/s494K ctx
dense
DeepSeek R1 0528 Qwen3 8B
C478B11.5 GB127 tok/s639K ctx
dense
DeepSeek R1 0528 Qwen3 8B
C478B11.5 GB127 tok/s639K ctx
dense
Qwen3 8B DeepSeek v3.2 Speciale Distill
C478B11.5 GB127 tok/s639K ctx
dense
Hermes 3 Llama 3.1 8B
C478B11.5 GB127 tok/s639K ctx
dense
Llama 2 7B Chat
C477B10.8 GB112 tok/s742K ctx
dense
Gemma 2 2B2B8.5 GB32 tok/s8K ctx
dense
Yi 9B Coder i1
C479B12.2 GB113 tok/s558K ctx
dense
Mistral 7B Instruct v0.2
C477B10.8 GB112 tok/s742K ctx
dense
Hermes 2 Pro Llama 3 8B
C478B11.5 GB127 tok/s639K ctx
dense
Dolphin3.0 Llama3.1 8B
C478B11.5 GB127 tok/s639K ctx
dense
Mistral 7B Instruct v0.3
C477B10.8 GB112 tok/s742K ctx
dense
granite 8b code instruct 4k
C478B11.5 GB127 tok/s639K ctx
dense
HelpingAI2 9B
C479B12.2 GB113 tok/s558K ctx
dense
dolphin 2.9.4 llama3.1 8b
C478B11.5 GB127 tok/s639K ctx
dense
HelpingAI2 9B i1
C479B12.2 GB113 tok/s558K ctx
dense
openchat 3.6 8b 20240522 IMat
C478B11.5 GB127 tok/s639K ctx
dense
HelpingAI 9B 200k i1
C479B12.2 GB113 tok/s558K ctx
dense
aya expanse 8b
C478B11.5 GB127 tok/s639K ctx
dense
HelpingAI 9B i1
C479B12.2 GB113 tok/s558K ctx
dense
EXAONE 3.5 7.8B Instruct
C477.8B11.4 GB125 tok/s657K ctx
dense
dolphin v2 8b abliterated i1
C478B11.5 GB127 tok/s639K ctx
dense
EXAONE 3.5 7.8B Instruct
C477.8B11.4 GB125 tok/s657K ctx
dense
aya expanse 8b orthogonal heretic i1
C478B11.5 GB127 tok/s639K ctx
dense
EXAONE 3.5 7.8B Instruct i1
C477.8B11.4 GB125 tok/s657K ctx
dense
Mistral 7B Instruct v0.3
C477B10.8 GB112 tok/s742K ctx
dense
Hermes 2 Pro Mistral 7B
C477B10.8 GB112 tok/s742K ctx
dense
exaone 3.0 7.8b it
C477.8B11.4 GB125 tok/s657K ctx
dense
zephyr 7B beta
C477B10.8 GB112 tok/s742K ctx
dense
Nous Hermes 2 Mistral 7B DPO
C477B10.8 GB112 tok/s742K ctx
dense
aya expanse 8b orthogonal heretic
C478B11.5 GB127 tok/s639K ctx
dense
Falcon H1R 7B
C477B10.8 GB112 tok/s742K ctx
dense
falcon mamba 7b instruct Q4 K M
C477B10.8 GB112 tok/s742K ctx
dense
zephyr 7B alpha
C477B10.8 GB112 tok/s742K ctx
dense
speechless zephyr code functionary 7b
C477B10.8 GB112 tok/s742K ctx
dense
internlm2 5 1 8b chat i1
C478B11.5 GB127 tok/s639K ctx
dense
internlm3 8b instruct abliterated i1
C478B11.5 GB127 tok/s639K ctx
dense
StarCoder2 7B
C467B10.8 GB112 tok/s742K ctx
dense
japanese stablelm instruct gamma 7B
C467B10.8 GB112 tok/s742K ctx
dense
Yi 1.5 6B Chat
C466B10.1 GB96 tok/s879K ctx
dense
Neural Chat 7B7B11.9 GB112 tok/s8K ctx
dense
Falcon H1 7B Instruct
C467B10.8 GB112 tok/s742K ctx
dense
Falcon H1R 7B7B10.8 GB112 tok/s742K ctx
dense
stablelm 2 zephyr 1 6b
C466B10.1 GB96 tok/s879K ctx
dense
Mamba Codestral 7B v0.1
C467B10.8 GB112 tok/s742K ctx
dense
stabilityai japanese stablelm base gamma 7b
C467B10.8 GB112 tok/s742K ctx
dense
baichuan2 7b chat
C467B10.8 GB112 tok/s742K ctx
dense
DiscoPOP zephyr 7b gemma
C467B10.8 GB112 tok/s742K ctx
dense
Yi 1.5 6B6B10.3 GB96 tok/s4K ctx
dense
starcoder2 7b
C467B10.8 GB112 tok/s742K ctx
dense
zephyr 7b beta Mistral 7B Instruct v0.2
C467B10.8 GB112 tok/s742K ctx
dense
OpenChat 3.5 7B Qwen v2.0 i1
C467B10.8 GB112 tok/s742K ctx
dense
OpenChat 3.5 7B Starling v2.0 i1
C467B10.8 GB112 tok/s742K ctx
dense
internlm2 math plus 7b IMat
C467B10.8 GB112 tok/s742K ctx
dense
CodeNinja 1.0 OpenChat 7B i1
C467B10.8 GB112 tok/s742K ctx
dense
Yi 1.5 6B Chat
C466B10.1 GB96 tok/s879K ctx
dense
internlm2 5 7b chat i1
C467B10.8 GB112 tok/s742K ctx
dense
stablelm 2 1 6b chat imatrix
C466B10.1 GB96 tok/s879K ctx
dense
OpenSafetyLab MD Judge v0 2 internlm2 7b
C467B10.8 GB112 tok/s742K ctx
dense
MD Judge v0 2 internlm2 7b i1
C467B10.8 GB112 tok/s742K ctx
dense
zephyr 7b gemma sft african ultrachat 100k
C467B10.8 GB112 tok/s742K ctx
dense
jointpreferences mistral 7b sft helpful
C467B10.8 GB112 tok/s742K ctx
dense
zephyr 7b dpo full i1
C467B10.8 GB112 tok/s742K ctx
dense
blossom v3 baichuan2 7b i1
C467B10.8 GB112 tok/s742K ctx
dense
blossom v1 baichuan 7b i1
C467B10.8 GB112 tok/s742K ctx
dense
Nemotron Mini 4B4B10.1 GB64 tok/s4K ctx
dense
HelpingAI2 6B
C466B10.1 GB96 tok/s879K ctx
dense
HelpingAI2 6B i1
C466B10.1 GB96 tok/s879K ctx
dense
Qwen 2.5 Coder 0.5B0.5B6.2 GB8 tok/s131K ctx
dense
StarCoder2 7B7B10.5 GB112 tok/s16K ctx
dense
DeepSeek LLM 67B67B52.4 GB6 tok/s4K ctx
dense
HelpingAI2.5 5B i1
C455B9.3 GB80 tok/s1072K ctx
dense
Qwen3.5 4B4B8.6 GB64 tok/s1361K ctx
dense
gemma 3 4b it
C454B8.6 GB64 tok/s1361K ctx
dense
gemma 3 4b it
C454B8.6 GB64 tok/s1361K ctx
dense
Qwen 3 0.6B0.6B6.9 GB10 tok/s33K ctx
dense
Llama 3.2 1B1B6.8 GB16 tok/s128K ctx
dense
Llama 3.2 3B Instruct
C443B8.2 GB48 tok/s1827K ctx
dense
Qwen2.5 3B Instruct3B7.9 GB48 tok/s1842K ctx
dense
Llama 3.2 3B Instruct
C433B7.9 GB48 tok/s1842K ctx
dense
Ministral 3 3B Instruct 2512
C433B7.9 GB48 tok/s1842K ctx
dense
HELVETE 3B
C433B7.9 GB48 tok/s1842K ctx
dense
Hermes 3 Llama 3.2 3B
C433B7.9 GB48 tok/s1842K ctx
dense
stablelm zephyr 3b
C433B7.9 GB48 tok/s1842K ctx
dense
StarCoder2 3B
C433B7.9 GB48 tok/s1842K ctx
dense
AI21 Jamba Reasoning 3B
C433B7.9 GB48 tok/s1842K ctx
dense
stablelm 3b 4e1t
C433B7.9 GB48 tok/s1842K ctx
dense
ai21labs AI21 Jamba Reasoning 3B
C433B7.9 GB48 tok/s1842K ctx
dense
ai21labs AI21 Jamba2 3B
C433B7.9 GB48 tok/s1842K ctx
dense
HelpingAI 3B hindi i1
C433B7.9 GB48 tok/s1842K ctx
dense
AI21 Jamba2 3B
C433B7.9 GB48 tok/s1842K ctx
dense
HelpingAI 3B hindi
C433B7.9 GB48 tok/s1842K ctx
dense
AI21 Jamba2 3B i1
C433B7.9 GB48 tok/s1842K ctx
dense
gemma 2 2b it
C422B7.6 GB32 tok/s2776K ctx
dense
EXAONE 3.5 2.4B Instruct
C422.4B7.4 GB38 tok/s2323K ctx
dense
gemma 2b2B7.2 GB32 tok/s2804K ctx
dense
Gemmasutra Mini 2B v1
C422B7.2 GB32 tok/s2804K ctx
dense
StarCoder2 3B3B8.0 GB48 tok/s16K ctx
dense
gemma 2 2b it
C422B7.2 GB32 tok/s2804K ctx
dense
Qwen 3.5 0.6B0.6B6.9 GB10 tok/s131K ctx
dense
Qwen2.5 1.5B Instruct1.5B6.8 GB24 tok/s3767K ctx
dense
DeepSeek R1 Distill Qwen 1.5B1.5B6.8 GB24 tok/s3767K ctx
dense
Yi Coder 1.5B Chat
C411.5B6.8 GB24 tok/s3767K ctx
dense
stablelm 2 zephyr 1.6b
C401.6B6.9 GB26 tok/s3526K ctx
dense
logos16v2 stablelm2 1.6b i1
C401.6B6.9 GB26 tok/s3526K ctx
dense
Yi Coder 1.5B
C401.5B6.8 GB24 tok/s3767K ctx
dense
Falcon H1 1.5B Instruct1.5B6.8 GB24 tok/s3767K ctx
dense
F
F Tier -- Can't Run (54 models)
Too heavy for this hardware.
TinyLlama 1.1B Chat v1.0
D401.1B6.5 GB18 tok/s5167K ctx
dense
EXAONE 4.0 1.2B
D401.2B6.6 GB19 tok/s4730K ctx
dense
Qwen 2.5 0.5B0.5B6.2 GB8 tok/s131K ctx
dense
Llama 3.2 1B Instruct Q8 0
D401B6.6 GB16 tok/s5663K ctx
dense
TinyLlama 1.1B Chat v0.3
D391.1B6.5 GB18 tok/s5167K ctx
dense
TinyLlama 1.1B Chat v0.61.1B6.5 GB18 tok/s5167K ctx
dense
Llama 3.2 1B Instruct
D391B6.4 GB16 tok/s5692K ctx
dense
gemma 3 1b it
D391B6.4 GB16 tok/s5692K ctx
dense
TinyLlama 1.1B Chat v1.0 imatrix
D391.1B6.5 GB18 tok/s5167K ctx
dense
Falcon3 1B Instruct abliterated
D391B6.4 GB16 tok/s5692K ctx
dense
SmolVLM 500M Instruct
D370.5B6.2 GB8 tok/s6702K ctx
dense
embeddinggemma 300M
D360.3B6.0 GB5 tok/s6729K ctx
dense
Solar Open 69B REAP i1
D3569B55.9 GB5 tok/s4K ctx
dense
granite embedding 107m multilingual
D330.11B5.9 GB2 tok/s6758K ctx
dense
Falcon H1 Tiny 90M Instruct
D330.09B5.9 GB2 tok/s6759K ctx
dense
Llama 3.3 70B Instruct
D2970B56.6 GB5 tok/s4K ctx
dense
stabilityai japanese stablelm instruct beta 70b
D2870B56.6 GB5 tok/s4K ctx
dense
Qwen 3 235B A22B235B151.9 GB2 tok/s4K ctx
moe
Qwen3-Coder 480B A35B Instruct480B301.4 GB2 tok/s4K ctx
moe
Qwen 3.5 122B A10B122B82.6 GB7 tok/s4K ctx
moe
Qwen 3.5 397B A17B397B250.7 GB2 tok/s4K ctx
moe
Mixtral 8x22B141B95.1 GB3 tok/s4K ctx
moe
Devstral 2 123B Instruct123B86.1 GB2 tok/s4K ctx
dense
Mistral Small 4 119B119B83.7 GB7 tok/s4K ctx
moe
Leanstral 119B A6B119B87.1 GB6 tok/s4K ctx
moe
Mistral Large 3675B422.8 GB2 tok/s4K ctx
moe
GLM-5744B478.6 GB2 tok/s4K ctx
moe
GLM-5.1754B484.7 GB2 tok/s4K ctx
moe
Kimi K2.51000B623.1 GB2 tok/s4K ctx
moe
Kimi K2.61000B623.1 GB2 tok/s4K ctx
moe
DeepSeek V3 671B671B474.6 GB2 tok/s4K ctx
moe
DeepSeek V3.1 671B671B474.6 GB2 tok/s4K ctx
moe
DeepSeek V4 Pro1600B869.6 GB2 tok/s4K ctx
moe
DeepSeek V4 Flash284B165.0 GB2 tok/s4K ctx
moe
DeepSeek R1 671B671B474.6 GB2 tok/s4K ctx
moe
DeepSeek Coder V2 236B236B208.3 GB2 tok/s4K ctx
moe
DeepSeek V2.5 236B236B208.3 GB2 tok/s4K ctx
moe
Llama 4 Scout 17B 16E109B75.1 GB7 tok/s4K ctx
moe
Llama 4 Maverick 17B 128E400B252.6 GB2 tok/s4K ctx
moe
Command R+ 104B104B72.6 GB2 tok/s4K ctx
dense
Command A 111B111B77.3 GB2 tok/s4K ctx
dense
Pixtral Large 124B124B86.7 GB2 tok/s4K ctx
dense
Llama 3.1 405B405B260.4 GB2 tok/s4K ctx
dense
GPT-OSS 120B117B82.0 GB2 tok/s4K ctx
dense
DeepSeek V3.2671B415.5 GB2 tok/s4K ctx
moe
MiniMax M2.7
F0230B149.8 GB2 tok/s4K ctx
moe
Solar Open 100B
F0100B78.4 GB2 tok/s4K ctx
dense
K EXAONE 236B A23B
F0236B177.3 GB2 tok/s4K ctx
dense
Baichuan M3 235B
F0235B176.6 GB2 tok/s4K ctx
dense
Baichuan M3 235B i1
F0235B176.6 GB2 tok/s4K ctx
dense
BaichuanMed OCR 72B i1
F072B58.1 GB4 tok/s4K ctx
dense
Solar Open 100B i1
F0100B78.4 GB2 tok/s4K ctx
dense
Qwen3.5 122B A10B122B79.8 GB2 tok/s4K ctx
dense
Qwen3.5 397B A17B397B294.4 GB2 tok/s4K ctx
dense
Compare with...
See How Other Hardware Stacks Up
Check tier lists for similar GPUs and Apple Silicon configs