AI Compatibility Tier List
RTX 4500 Ada 24GB
374 AI models ranked by how well they run on your 24 GB VRAM hardware. Best match: Qwen3-Coder 30B A3B Instruct (score: 95).
374
Models tested
295
Can run
24 GB
VRAM
52
Best tok/s
Tier Distribution
S 21
A 58
B 39
C 177
F 79
S: 21A: 58B: 39C: 177F: 79
S
S Tier -- Perfect Fit (21 models)
These models run flawlessly with headroom to spare.
S95
30.5B23.4 GB52 tok/s23K ctx
moe
S94
30B23.1 GB53 tok/s26K ctx
moe
S94
21B18.6 GB66 tok/s52K ctx
moe
S93
14B14.3 GB43 tok/s80K ctx
dense
S93
14.7B15.3 GB41 tok/s33K ctx
dense
S93
30.5B23.4 GB52 tok/s23K ctx
moe
S92
9B11.0 GB67 tok/s111K ctx
dense
S92
27B22.9 GB22 tok/s21K ctx
dense
S91
24B20.4 GB25 tok/s40K ctx
dense
S91
24B20.4 GB25 tok/s40K ctx
dense
S91
27B20.7 GB17 tok/s69K ctx
+1dense
S90
8B10.4 GB75 tok/s115K ctx
dense
S89
24B20.4 GB25 tok/s40K ctx
dense
S89
30B24.5 GB38 tok/s13K ctx
moe
S88
30B24.0 GB15 tok/s16K ctx
dense
S87
4B7.9 GB56 tok/s131K ctx
dense
S87
25.2B22.3 GB55 tok/s23K ctx
moe
S87
14B14.3 GB43 tok/s80K ctx
multimodal
S87
22B19.2 GB24 tok/s48K ctx
dense
S87
19B17.3 GB32 tok/s8K ctx
dense
S85
8B10.1 GB75 tok/s130K ctx
dense
A
A Tier -- Great Fit (58 models)
Excellent performance with comfortable memory margins.
A85
16B16.4 GB83 tok/s53K ctx
moe
A84
14B14.9 GB43 tok/s16K ctx
dense
A84
24B20.4 GB25 tok/s40K ctx
vision
A84
3.8B7.1 GB53 tok/s131K ctx
dense
A83
14B14.8 GB43 tok/s66K ctx
dense
A83
24B20.4 GB25 tok/s40K ctx
dense
A83
20B18.7 GB30 tok/s8K ctx
dense
A83
8B10.1 GB75 tok/s8K ctx
dense
A82
12B15.5 GB37 tok/s44K ctx
dense
A82
35B26.1 GB30 tok/s4K ctx
moe
A82
24B20.4 GB25 tok/s33K ctx
dense
A81
8B10.4 GB75 tok/s115K ctx
multimodal
A81
24B20.4 GB25 tok/s40K ctx
dense
A81
8B10.1 GB75 tok/s2K ctx
dense
A80
9B11.2 GB67 tok/s100K ctx
dense
A80
7B9.5 GB86 tok/s8K ctx
dense
A79
7B8.4 GB87 tok/s33K ctx
dense
A79
13B13.7 GB47 tok/s33K ctx
dense
A79
4B7.9 GB56 tok/s33K ctx
dense
A79
7B15.4 GB80 tok/s16K ctx
dense
A78
32B26.7 GB11 tok/s5K ctx
dense
A78
7B14.9 GB80 tok/s8K ctx
dense
A78
7B9.5 GB86 tok/s8K ctx
dense
A77
14B14.8 GB43 tok/s33K ctx
dense
A77
9B9.4 GB68 tok/s131K ctx
dense
A77
0.57B6.4 GB8 tok/s8K ctx
dense
A77
13B23.4 GB43 tok/s16K ctx
dense
A76
8B9.5 GB57 tok/s128K ctx
dense
A76
7B15.4 GB80 tok/s8K ctx
dense
A76
7B15.4 GB80 tok/s8K ctx
dense
A76
35B28.8 GB22 tok/s4K ctx
+1moe
A75
8B10.1 GB75 tok/s8K ctx
dense
A75
7B9.5 GB86 tok/s8K ctx
dense
A75
12B13.1 GB50 tok/s88K ctx
dense
A75
7B8.4 GB87 tok/s131K ctx
dense
A75
0.57B5.6 GB8 tok/s8K ctx
dense
A75
8B10.6 GB75 tok/s104K ctx
dense
A75
13B23.4 GB43 tok/s4K ctx
dense
A74
7B8.1 GB92 tok/s262K ctx
state-space
A73
9B21.0 GB62 tok/s16K ctx
dense
A73
7B15.4 GB80 tok/s4K ctx
dense
A72
32B26.7 GB11 tok/s5K ctx
dense
A72
13B23.4 GB43 tok/s8K ctx
dense
A72
13B23.4 GB43 tok/s4K ctx
dense
A72
7B10.5 GB86 tok/s8K ctx
dense
A72
0.34B5.5 GB5 tok/s1K ctx
dense
A72
3B7.3 GB42 tok/s131K ctx
dense
A71
7B15.4 GB80 tok/s8K ctx
dense
A71
7B15.4 GB80 tok/s8K ctx
dense
A71
8B10.1 GB75 tok/s128K ctx
dense
A71
7B9.5 GB86 tok/s4K ctx
dense
A71
7B9.5 GB86 tok/s4K ctx
dense
A71
0.34B5.5 GB5 tok/s1K ctx
dense
A71
32B26.7 GB11 tok/s5K ctx
dense
A71
9B9.4 GB68 tok/s128K ctx
dense
A70
13B24.9 GB26 tok/s15K ctx
dense
A70
5.1B6.9 GB71 tok/s128K ctx
dense
A70
30B25.5 GB13 tok/s10K ctx
dense
B
B Tier -- Good Fit (39 models)
Solid performance, may need quantization or reduced context.
B70
32B26.7 GB11 tok/s4K ctx
dense
B69
7B8.4 GB87 tok/s131K ctx
dense
B69
0.34B5.5 GB5 tok/s1K ctx
dense
B69
4B7.8 GB56 tok/s128K ctx
dense
B67
3.8B11.5 GB53 tok/s50K ctx
dense
B67
4B7.2 GB56 tok/s128K ctx
dense
B67
13B22.4 GB37 tok/s19K ctx
dense
B67
3B5.9 GB42 tok/s262K ctx
multimodal
B67
14B14.8 GB43 tok/s66K ctx
dense
B67
8B10.1 GB75 tok/s33K ctx
dense
B67
9B13.9 GB50 tok/s8K ctx
dense
B67
0.14B4.1 GB2 tok/s8K ctx
dense
B66
4B11.6 GB56 tok/s50K ctx
dense
B66
7B9.5 GB86 tok/s4K ctx
dense
B66
7B8.4 GB87 tok/s33K ctx
dense
B66
13B24.9 GB26 tok/s8K ctx
dense
B66
2B6.2 GB28 tok/s131K ctx
dense
B65
11B12.0 GB55 tok/s16K ctx
vision
B65
32B26.7 GB11 tok/s5K ctx
dense
B65
7B7.7 GB88 tok/s8K ctx
dense
B65
3B7.3 GB42 tok/s131K ctx
dense
B64
12B13.1 GB50 tok/s88K ctx
dense
B64
14B14.9 GB43 tok/s64K ctx
dense
B64
34B27.7 GB10 tok/s4K ctx
dense
B63
35B27.1 GB10 tok/s4K ctx
dense
B63
22B19.2 GB26 tok/s33K ctx
dense
B63
1.7B6.0 GB24 tok/s33K ctx
dense
B63
32B26.7 GB11 tok/s5K ctx
dense
B62
7B9.5 GB86 tok/s8K ctx
dense
B62
9B10.3 GB68 tok/s131K ctx
dense
B62
3B7.6 GB42 tok/s8K ctx
dense
B62
3B6.4 GB42 tok/s131K ctx
dense
B61
0.02B3.6 GB2 tok/s0K ctx
dense
B60
1.5B4.6 GB21 tok/s33K ctx
dense
B60
8B10.4 GB75 tok/s115K ctx
dense
B58
3B6.8 GB42 tok/s128K ctx
dense
B57
15B28.7 GB17 tok/s8K ctx
dense
B56
7B15.4 GB80 tok/s4K ctx
dense
B55
8B10.1 GB87 tok/s128K ctx
state-space
C
C Tier -- Tight Fit (177 models)
Runs with offloading or significant compromises.
C55
9B10.3 GB68 tok/s4K ctx
dense
C54
15B15.3 GB35 tok/s16K ctx
dense
C53
3B7.1 GB42 tok/s128K ctx
dense
C53
7B8.4 GB87 tok/s4K ctx
dense
C53
7B14.9 GB80 tok/s4K ctx
dense
C53
32B25.3 GB13 tok/s8K ctx
dense
Codestral 22B v0.1
C5322B19.3 GB25 tok/s45K ctx
dense
Codestral RAG 19B Pruned i1
C5319B17.1 GB29 tok/s65K ctx
dense
Codestral 22B v0.1
C5322B19.3 GB25 tok/s45K ctx
dense
internlm2 5 20b chat
C5320B17.8 GB28 tok/s58K ctx
dense
internlm2 limarp chat 20b
C5320B17.8 GB28 tok/s58K ctx
dense
internlm2 math plus 20b i1
C5320B17.8 GB28 tok/s58K ctx
dense
Codestral 22B v0.1
C5222B19.3 GB25 tok/s45K ctx
dense
Codestral 21B Pruned i1
C5221B18.6 GB27 tok/s51K ctx
dense
Codestral 22B v0.1 IMat
C5222B19.3 GB25 tok/s45K ctx
dense
Codestral 22B v0.1 i1
C5222B19.3 GB25 tok/s45K ctx
dense
C52
7B9.5 GB86 tok/s8K ctx
dense
C52
1.5B4.6 GB21 tok/s33K ctx
dense
C52
8B10.1 GB75 tok/s8K ctx
dense
C52
1.1B4.3 GB15 tok/s4K ctx
dense
Phi 4 reasoning vision 15B
C5215B14.2 GB37 tok/s105K ctx
dense
StarCoder2 15B
C5215B14.2 GB37 tok/s105K ctx
dense
C51
14B13.5 GB40 tok/s119K ctx
dense
starcoder2 15b instruct v0.1
C5115B14.2 GB37 tok/s105K ctx
dense
starcoder2 15b i1
C5115B14.2 GB37 tok/s105K ctx
dense
starcoder2 15b instruct v0.1
C5115B14.2 GB37 tok/s105K ctx
dense
HelpingAI 15B i1
C5115B14.2 GB37 tok/s105K ctx
dense
NousResearch Hermes 4 14B
C5114B13.5 GB40 tok/s119K ctx
dense
C51
7B9.5 GB86 tok/s8K ctx
dense
C51
8B10.1 GB75 tok/s33K ctx
dense
GGUF SOLARized GraniStral 14B 2102 YeAM HCT 32QKV
C5114B13.5 GB40 tok/s119K ctx
dense
internlm JanusCoder 14B
C5114B13.5 GB40 tok/s119K ctx
dense
GGUF SOLARized GraniStral 14B 1902 YeAM HCT
C5114B13.5 GB40 tok/s119K ctx
dense
C51
7B9.5 GB86 tok/s33K ctx
dense
gemma 3 12b it
C5112B12.0 GB47 tok/s152K ctx
dense
C50
7B9.5 GB86 tok/s8K ctx
dense
SOLAR 10.7B Instruct v1.0 uncensored
C5010.7B11.1 GB52 tok/s181K ctx
dense
Nous Hermes 2 SOLAR 10.7B
C5010.7B11.1 GB52 tok/s181K ctx
dense
C50
9B9.8 GB62 tok/s231K ctx
dense
Qwen3.5 9B Uncensored HauhauCS Aggressive
C509B9.8 GB62 tok/s231K ctx
dense
C50
12B24.1 GB27 tok/s4K ctx
dense
solar finalised finetuned Model 10.7B i1
C5010.7B11.1 GB52 tok/s181K ctx
dense
Qwen3.5 9B
C509B9.8 GB62 tok/s231K ctx
dense
C50
24B20.8 GB23 tok/s34K ctx
dense
SOLAR 10.7B v1.0
C5010.7B11.1 GB52 tok/s181K ctx
dense
cognitivecomputations Dolphin Mistral 24B Venice Edition
C5024B20.8 GB23 tok/s34K ctx
dense
Mistral Small 24B Instruct 2501
C5024B20.8 GB23 tok/s34K ctx
dense
Meta Llama 3.1 8B Instruct
C498B9.1 GB70 tok/s270K ctx
dense
C49
27B22.9 GB21 tok/s21K ctx
dense
mistral small 3.1 24b instruct 2503 hf
C4924B20.8 GB23 tok/s34K ctx
dense
llava llama 3 8b v1 1
C498B9.1 GB70 tok/s270K ctx
dense
cognitivecomputations Dolphin3.0 R1 Mistral 24B
C4924B20.8 GB23 tok/s34K ctx
dense
Yi Coder 9B Chat
C499B9.8 GB62 tok/s231K ctx
dense
C49
7B9.5 GB86 tok/s8K ctx
dense
glm 4 9b chat 1m
C499B9.8 GB62 tok/s231K ctx
dense
C49
34B27.7 GB10 tok/s4K ctx
dense
C49
1.5B4.6 GB21 tok/s131K ctx
dense
C49
8B9.1 GB70 tok/s270K ctx
dense
Dolphin Mistral GLM 4.7 Flash 24B Venice Edition Thinking Uncensored i1
C4924B20.8 GB23 tok/s34K ctx
dense
Meta Llama 3 8B Instruct
C498B9.1 GB70 tok/s270K ctx
dense
C49
8B9.1 GB70 tok/s270K ctx
dense
C49
27B22.9 GB21 tok/s21K ctx
dense
Dolphin3.0 Llama3.1 8B
C498B9.1 GB70 tok/s270K ctx
dense
Llama 3 8B Instruct 32k v0.1
C498B9.1 GB70 tok/s270K ctx
dense
Meta Llama 3.1 8B Instruct
C498B9.1 GB70 tok/s270K ctx
dense
vntl llama3 8b v2
C498B9.1 GB70 tok/s270K ctx
dense
Yi 1.5 9B Chat
C499B9.8 GB62 tok/s231K ctx
dense
DeepSeek R1 0528 Qwen3 8B
C498B9.1 GB70 tok/s270K ctx
dense
Llama 2 7B Chat
C497B8.4 GB80 tok/s320K ctx
dense
Helply 10.2b chat i1
C4910.2B10.7 GB55 tok/s194K ctx
dense
HelpingAI2.5 10B i1
C4910B10.6 GB56 tok/s199K ctx
dense
falcon mamba 7b instruct Q4 K M
C497B8.4 GB92 tok/s320K ctx
dense
Mistral 7B Instruct v0.2
C497B8.4 GB80 tok/s320K ctx
dense
DeepSeek R1 0528 Qwen3 8B
C498B9.1 GB70 tok/s270K ctx
dense
Mistral 7B Instruct v0.3
C497B8.4 GB80 tok/s320K ctx
dense
Qwen3 8B DeepSeek v3.2 Speciale Distill
C498B9.1 GB70 tok/s270K ctx
dense
Hermes 3 Llama 3.1 8B
C498B9.1 GB70 tok/s270K ctx
dense
gemma 3 27b it
C4927B22.9 GB21 tok/s21K ctx
dense
cognitivecomputations Dolphin Mistral 24B Venice Edition
C4924B20.8 GB23 tok/s34K ctx
dense
Hermes 2 Pro Llama 3 8B
C498B9.1 GB70 tok/s270K ctx
dense
Dolphin3.0 Llama3.1 8B
C498B9.1 GB70 tok/s270K ctx
dense
granite 8b code instruct 4k
C498B9.1 GB70 tok/s270K ctx
dense
Yi 9B Coder i1
C499B9.8 GB62 tok/s231K ctx
dense
dolphin 2.9.4 llama3.1 8b
C498B9.1 GB70 tok/s270K ctx
dense
Mamba Codestral 7B v0.1
C497B8.4 GB92 tok/s320K ctx
dense
HelpingAI2 9B
C499B9.8 GB62 tok/s231K ctx
dense
openchat 3.6 8b 20240522 IMat
C498B9.1 GB70 tok/s270K ctx
dense
Mistral 7B Instruct v0.3
C497B8.4 GB80 tok/s320K ctx
dense
Hermes 2 Pro Mistral 7B
C497B8.4 GB80 tok/s320K ctx
dense
EXAONE 3.5 7.8B Instruct
C497.8B9.0 GB72 tok/s279K ctx
dense
aya expanse 8b
C498B9.1 GB70 tok/s270K ctx
dense
zephyr 7B beta
C497B8.4 GB80 tok/s320K ctx
dense
HelpingAI2 9B i1
C499B9.8 GB62 tok/s231K ctx
dense
EXAONE 3.5 7.8B Instruct
C497.8B9.0 GB72 tok/s279K ctx
dense
dolphin v2 8b abliterated i1
C498B9.1 GB70 tok/s270K ctx
dense
HelpingAI 9B 200k i1
C499B9.8 GB62 tok/s231K ctx
dense
Nous Hermes 2 Mistral 7B DPO
C497B8.4 GB80 tok/s320K ctx
dense
EXAONE 3.5 7.8B Instruct i1
C497.8B9.0 GB72 tok/s279K ctx
dense
HelpingAI 9B i1
C499B9.8 GB62 tok/s231K ctx
dense
aya expanse 8b orthogonal heretic i1
C498B9.1 GB70 tok/s270K ctx
dense
Falcon H1R 7B
C497B8.4 GB80 tok/s320K ctx
dense
exaone 3.0 7.8b it
C497.8B9.0 GB72 tok/s279K ctx
dense
zephyr 7B alpha
C497B8.4 GB80 tok/s320K ctx
dense
speechless zephyr code functionary 7b
C497B8.4 GB80 tok/s320K ctx
dense
StarCoder2 7B
C497B8.4 GB80 tok/s320K ctx
dense
aya expanse 8b orthogonal heretic
C498B9.1 GB70 tok/s270K ctx
dense
japanese stablelm instruct gamma 7B
C497B8.4 GB80 tok/s320K ctx
dense
Falcon H1 7B Instruct
C497B8.4 GB80 tok/s320K ctx
dense
C49
7B8.4 GB80 tok/s320K ctx
dense
internlm2 5 1 8b chat i1
C498B9.1 GB70 tok/s270K ctx
dense
internlm3 8b instruct abliterated i1
C498B9.1 GB70 tok/s270K ctx
dense
stabilityai japanese stablelm base gamma 7b
C497B8.4 GB80 tok/s320K ctx
dense
Yi 1.5 6B Chat
C496B7.7 GB84 tok/s388K ctx
dense
baichuan2 7b chat
C497B8.4 GB80 tok/s320K ctx
dense
DiscoPOP zephyr 7b gemma
C497B8.4 GB80 tok/s320K ctx
dense
starcoder2 7b
C497B8.4 GB80 tok/s320K ctx
dense
zephyr 7b beta Mistral 7B Instruct v0.2
C487B8.4 GB80 tok/s320K ctx
dense
OpenChat 3.5 7B Qwen v2.0 i1
C487B8.4 GB80 tok/s320K ctx
dense
OpenChat 3.5 7B Starling v2.0 i1
C487B8.4 GB80 tok/s320K ctx
dense
internlm2 math plus 7b IMat
C487B8.4 GB80 tok/s320K ctx
dense
stablelm 2 zephyr 1 6b
C486B7.7 GB84 tok/s388K ctx
dense
CodeNinja 1.0 OpenChat 7B i1
C487B8.4 GB80 tok/s320K ctx
dense
C48
6B7.9 GB84 tok/s4K ctx
dense
internlm2 5 7b chat i1
C487B8.4 GB80 tok/s320K ctx
dense
OpenSafetyLab MD Judge v0 2 internlm2 7b
C487B8.4 GB80 tok/s320K ctx
dense
MD Judge v0 2 internlm2 7b i1
C487B8.4 GB80 tok/s320K ctx
dense
zephyr 7b gemma sft african ultrachat 100k
C487B8.4 GB80 tok/s320K ctx
dense
jointpreferences mistral 7b sft helpful
C487B8.4 GB80 tok/s320K ctx
dense
zephyr 7b dpo full i1
C487B8.4 GB80 tok/s320K ctx
dense
blossom v3 baichuan2 7b i1
C487B8.4 GB80 tok/s320K ctx
dense
blossom v1 baichuan 7b i1
C487B8.4 GB80 tok/s320K ctx
dense
Yi 1.5 6B Chat
C486B7.7 GB84 tok/s388K ctx
dense
stablelm 2 1 6b chat imatrix
C486B7.7 GB84 tok/s388K ctx
dense
C48
2B6.1 GB28 tok/s8K ctx
dense
C48
7B8.1 GB87 tok/s16K ctx
dense
HelpingAI2 6B
C486B7.7 GB84 tok/s388K ctx
dense
HelpingAI2 6B i1
C486B7.7 GB84 tok/s388K ctx
dense
C48
4B7.7 GB56 tok/s4K ctx
dense
C48
1B4.3 GB14 tok/s33K ctx
dense
HelpingAI2.5 5B i1
C475B6.9 GB70 tok/s482K ctx
dense
C46
4B6.2 GB56 tok/s623K ctx
dense
C46
0.5B3.8 GB7 tok/s131K ctx
dense
gemma 3 4b it
C464B6.2 GB56 tok/s623K ctx
dense
gemma 3 4b it
C464B6.2 GB56 tok/s623K ctx
dense
Llama 3.2 3B Instruct
C453B5.8 GB42 tok/s844K ctx
dense
C45
0.6B4.5 GB8 tok/s33K ctx
dense
C45
3B5.5 GB42 tok/s859K ctx
dense
C44
1B4.4 GB14 tok/s128K ctx
dense
Llama 3.2 3B Instruct
C443B5.5 GB42 tok/s859K ctx
dense
Ministral 3 3B Instruct 2512
C443B5.5 GB42 tok/s859K ctx
dense
HELVETE 3B
C443B5.5 GB42 tok/s859K ctx
dense
Hermes 3 Llama 3.2 3B
C443B5.5 GB42 tok/s859K ctx
dense
stablelm zephyr 3b
C443B5.5 GB42 tok/s859K ctx
dense
StarCoder2 3B
C443B5.5 GB42 tok/s859K ctx
dense
AI21 Jamba Reasoning 3B
C443B5.5 GB42 tok/s859K ctx
dense
stablelm 3b 4e1t
C443B5.5 GB42 tok/s859K ctx
dense
ai21labs AI21 Jamba Reasoning 3B
C443B5.5 GB42 tok/s859K ctx
dense
ai21labs AI21 Jamba2 3B
C443B5.5 GB42 tok/s859K ctx
dense
HelpingAI 3B hindi i1
C443B5.5 GB42 tok/s859K ctx
dense
AI21 Jamba2 3B
C443B5.5 GB42 tok/s859K ctx
dense
HelpingAI 3B hindi
C443B5.5 GB42 tok/s859K ctx
dense
AI21 Jamba2 3B i1
C443B5.5 GB42 tok/s859K ctx
dense
gemma 2 2b it
C432B5.2 GB28 tok/s1301K ctx
dense
EXAONE 3.5 2.4B Instruct
C432.4B5.0 GB34 tok/s1094K ctx
dense
C43
3B5.6 GB42 tok/s16K ctx
dense
C43
2B4.8 GB28 tok/s1330K ctx
dense
Gemmasutra Mini 2B v1
C432B4.8 GB28 tok/s1330K ctx
dense
gemma 2 2b it
C422B4.8 GB28 tok/s1330K ctx
dense
C42
0.6B4.5 GB8 tok/s131K ctx
dense
C41
1.5B4.4 GB21 tok/s1801K ctx
dense
C41
1.5B4.4 GB21 tok/s1801K ctx
dense
Yi Coder 1.5B Chat
C411.5B4.4 GB21 tok/s1801K ctx
dense
stablelm 2 zephyr 1.6b
C411.6B4.5 GB22 tok/s1683K ctx
dense
logos16v2 stablelm2 1.6b i1
C411.6B4.5 GB22 tok/s1683K ctx
dense
Yi Coder 1.5B
C411.5B4.4 GB21 tok/s1801K ctx
dense
C41
1.5B4.4 GB21 tok/s1801K ctx
dense
TinyLlama 1.1B Chat v1.0
C401.1B4.1 GB15 tok/s2486K ctx
dense
F
F Tier -- Can't Run (79 models)
Too heavy for this hardware.
EXAONE 4.0 1.2B
D401.2B4.2 GB17 tok/s2272K ctx
dense
Llama 3.2 1B Instruct Q8 0
D401B4.2 GB14 tok/s2714K ctx
dense
TinyLlama 1.1B Chat v0.3
D401.1B4.1 GB15 tok/s2486K ctx
dense
D40
0.5B3.8 GB7 tok/s131K ctx
dense
D40
1.1B4.1 GB15 tok/s2486K ctx
dense
Llama 3.2 1B Instruct
D401B4.0 GB14 tok/s2743K ctx
dense
gemma 3 1b it
D401B4.0 GB14 tok/s2743K ctx
dense
TinyLlama 1.1B Chat v1.0 imatrix
D391.1B4.1 GB15 tok/s2486K ctx
dense
Falcon3 1B Instruct abliterated
D391B4.0 GB14 tok/s2743K ctx
dense
D38
34B27.7 GB10 tok/s4K ctx
dense
SmolVLM 500M Instruct
D370.5B3.8 GB7 tok/s3246K ctx
dense
EXAONE 4.0 32B
D3732B26.6 GB11 tok/s5K ctx
dense
aya expanse 32b heretic MPOA i1
D3732B26.6 GB11 tok/s5K ctx
dense
Baichuan M2 32B Q4 K M
D3732B26.6 GB11 tok/s5K ctx
dense
baichuan inc Baichuan M2 32B
D3732B26.6 GB11 tok/s5K ctx
dense
embeddinggemma 300M
D360.3B3.6 GB4 tok/s3273K ctx
dense
granite embedding 107m multilingual
D330.11B3.5 GB2 tok/s3302K ctx
dense
Falcon H1 Tiny 90M Instruct
D330.09B3.5 GB2 tok/s3303K ctx
dense
D31
35B28.8 GB8 tok/s4K ctx
dense
Qwen3.5 35B A3B
D3035B28.8 GB8 tok/s4K ctx
dense
F0
70B50.9 GB2 tok/s4K ctx
dense
F0
70B50.9 GB2 tok/s4K ctx
dense
F0
72B52.1 GB2 tok/s4K ctx
dense
F0
235B149.5 GB2 tok/s4K ctx
moe
F0
80B53.6 GB4 tok/s4K ctx
moe
F0
480B299.0 GB2 tok/s4K ctx
moe
F0
122B80.2 GB2 tok/s4K ctx
moe
F0
397B248.3 GB2 tok/s4K ctx
moe
F0
27B31.0 GB7 tok/s6K ctx
dense
F0
27B31.0 GB7 tok/s6K ctx
dense
F0
47B33.9 GB9 tok/s4K ctx
moe
F0
141B92.7 GB2 tok/s4K ctx
moe
F0
40B33.9 GB5 tok/s4K ctx
dense
F0
30B48.3 GB3 tok/s4K ctx
dense
F0
123B83.7 GB2 tok/s4K ctx
dense
F0
119B81.3 GB2 tok/s4K ctx
moe
F0
119B84.7 GB2 tok/s4K ctx
moe
F0
675B420.4 GB2 tok/s4K ctx
+1moe
F0
744B476.2 GB2 tok/s4K ctx
moe
F0
754B482.3 GB2 tok/s4K ctx
moe
F0
1000B620.7 GB2 tok/s4K ctx
moe
F0
1000B620.7 GB2 tok/s4K ctx
+1moe
F0
48B33.5 GB4 tok/s4K ctx
linear
F0
20B36.0 GB8 tok/s6K ctx
dense
F0
671B472.2 GB2 tok/s4K ctx
moe
F0
671B472.2 GB2 tok/s4K ctx
moe
F0
1600B867.2 GB2 tok/s4K ctx
moe
F0
284B162.6 GB2 tok/s4K ctx
moe
F0
671B472.2 GB2 tok/s4K ctx
moe
F0
236B205.9 GB2 tok/s4K ctx
moe
F0
236B205.9 GB2 tok/s4K ctx
moe
F0
67B50.0 GB2 tok/s4K ctx
dense
F0
109B72.7 GB2 tok/s4K ctx
moe
F0
400B250.2 GB2 tok/s4K ctx
moe
F0
104B70.2 GB2 tok/s4K ctx
dense
F0
111B74.9 GB2 tok/s4K ctx
dense
F0
124B84.3 GB2 tok/s4K ctx
dense
F0
72B52.1 GB2 tok/s4K ctx
dense
F0
72B52.1 GB2 tok/s4K ctx
dense
F0
70B50.9 GB2 tok/s4K ctx
dense
F0
405B258.0 GB2 tok/s4K ctx
dense
F0
70B50.9 GB2 tok/s4K ctx
dense
F0
117B79.6 GB2 tok/s4K ctx
dense
F0
30.7B36.7 GB5 tok/s4K ctx
dense
F0
671B413.1 GB2 tok/s4K ctx
moe
MiniMax M2.7
F0230B147.4 GB2 tok/s4K ctx
moe
Solar Open 100B
F0100B76.0 GB2 tok/s4K ctx
dense
Qwen3 48B A4B Savant Commander Distill 12X Closed Open Heretic Uncensored
F048B38.2 GB3 tok/s4K ctx
dense
K EXAONE 236B A23B
F0236B174.9 GB2 tok/s4K ctx
dense
Llama 3.3 70B Instruct
F070B54.2 GB2 tok/s4K ctx
dense
Baichuan M3 235B
F0235B174.2 GB2 tok/s4K ctx
dense
Baichuan M3 235B i1
F0235B174.2 GB2 tok/s4K ctx
dense
BaichuanMed OCR 72B i1
F072B55.7 GB2 tok/s4K ctx
dense
Solar Open 100B i1
F0100B76.0 GB2 tok/s4K ctx
dense
Solar Open 69B REAP i1
F069B53.5 GB2 tok/s4K ctx
dense
Hermes 4.3 36B
F036B29.5 GB8 tok/s4K ctx
dense
stabilityai japanese stablelm instruct beta 70b
F070B54.2 GB2 tok/s4K ctx
dense
F0
122B77.4 GB2 tok/s4K ctx
dense
F0
397B292.0 GB2 tok/s4K ctx
dense
Compare with...
See How Other Hardware Stacks Up
Check tier lists for similar GPUs and Apple Silicon configs
