मॉडल

Qwenimage-text-to-text253 स्पीड टेस्ट

Qwen3.6-35B-A3B

Qwen / Qwen3.6-35B-A3B

MoE3B active / 35B totalqwen3_5_moeconversational

सर्वश्रेष्ठ6.0k tok/s

मीडियन80.1 tok/s

न्यूनतम6.4 tok/s

Eval score· 2 evalsBF16+

91.6%

Qwenimage-text-to-text223 स्पीड टेस्ट

Qwen3.6-27B

Qwen / Qwen3.6-27B

27Bqwen3_5conversational

सर्वश्रेष्ठ1.4k tok/s

मीडियन46.0 tok/s

न्यूनतम1.6 tok/s

Qwenimage-text-to-text452 स्पीड टेस्ट

Qwen3.8-Flash-Next

Qwen / Qwen3.8-Flash-Next

180Bqwen4_expconversational

सर्वश्रेष्ठ231 tok/s

मीडियन56.0 tok/s

न्यूनतम1.9 tok/s

Eval score· 2 evalsW4B

85.7%

Qwen3.5-9B-Base

Qwen / Qwen3.5-9B-Base

9Bqwen3_5conversational

gemma-4-26B-A4B

google / gemma-4-26B-A4B

MoE4B active / 26B totalgemma4

Qwentext-generation26 स्पीड टेस्ट

Ornith-1.5-35B-A3B

ornith-ai / Ornith-1.5-35B-A3B

MoE3B active / 35B totalqwen3_5_moeimage-text-to-textconversational

सर्वश्रेष्ठ3.7k tok/s

मीडियन1.6k tok/s

न्यूनतम35.4 tok/s

Qwenimage-text-to-text236 स्पीड टेस्ट

Qwen3.8-27B-MTP-GGUF

Jackrong / Qwen3.8-27B-MTP-GGUF

27Bllama.cppvisionmultimodal

सर्वश्रेष्ठ72.6 tok/s

मीडियन36.6 tok/s

न्यूनतम11.6 tok/s

text-generation33 स्पीड टेस्ट

NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16

nvidia / NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16

MoE3B active / 30B totalnemotron_hnvidianemotron-3.5

सर्वश्रेष्ठ7.2k tok/s

मीडियन125 tok/s

न्यूनतम46.9 tok/s

Deepseektext-generation139 स्पीड टेस्ट

DeepSeek-V4-Flash-0731

deepseek-ai / DeepSeek-V4-Flash-0731

MoE304Bdeepseek_v48-bitfp8

सर्वश्रेष्ठ404 tok/s

मीडियन24.9 tok/s

न्यूनतम1.6 tok/s

Eval score· 5 evalsFP8

90.4%

image-text-to-text84 स्पीड टेस्ट

GLM-5.3-Flash

zai-org / GLM-5.3-Flash

321Bglm5_nextconversationalen

सर्वश्रेष्ठ1.5k tok/s

मीडियन60.8 tok/s

न्यूनतम3.7 tok/s

Eval score· 8 evalsNVFP4+

73.4%

Gemmaany-to-any1 स्पीड टेस्ट

gemma-4-12B

google / gemma-4-12B

12Bgemma4_unifiedimage-text-to-text

सर्वश्रेष्ठ25.8 tok/s

मीडियन25.8 tok/s

न्यूनतम25.8 tok/s

Qwen3.5-35B-A3B-Base

Qwen / Qwen3.5-35B-A3B-Base

MoE3B active / 35B totalqwen3_5_moeconversational

NVIDIA-Nemotron-Nano-12B-v2-Base

nvidia / NVIDIA-Nemotron-Nano-12B-v2-Base

12Bnvidiaenes

Minimaxtext-generation40 स्पीड टेस्ट

MiniMax-M2.7

MiniMaxAI / MiniMax-M2.7

text-generation129 स्पीड टेस्ट

Ornith-1.5-35B-A3B-GGUF

ornith-ai / Ornith-1.5-35B-A3B-GGUF

MoE3B active / 35B totalconversational

gemma-4-31B

google / gemma-4-31B

31Bgemma4

Gemmaany-to-any3 स्पीड टेस्ट

gemma-4-E4B

google / gemma-4-E4B

4Bgemma4image-text-to-text

सर्वश्रेष्ठ71.2 tok/s

मीडियन70.5 tok/s

न्यूनतम70.5 tok/s

Gpttext-generation101 स्पीड टेस्ट

gpt-oss-20b

openai / gpt-oss-20b

MoE20Bgpt_ossvllmconversational

सर्वश्रेष्ठ5.2k tok/s

मीडियन180 tok/s

न्यूनतम12.0 tok/s

Qwenimage-text-to-text92 स्पीड टेस्ट

Qwen3.5-122B-A10B

Qwen / Qwen3.5-122B-A10B

MoE10B active / 122B totalqwen3_5_moeconversational

granite-4.1-30b-base

ibm-granite / granite-4.1-30b-base

30Bgranitelanguagegranite-4.1

Deepseekimage-text-to-text100 स्पीड टेस्ट

DeepSeek-V4.1-Flash-1017

deepseek-ai / DeepSeek-V4.1-Flash-1017

763Bdeepseek_v41text-generation8-bit

Ornith-1.5-9B

ornith-ai / Ornith-1.5-9B

Qwentext-generation108 स्पीड टेस्ट

9Bqwen3_5image-text-to-textconversational

सर्वश्रेष्ठ3.8k tok/s

मीडियन75.3 tok/s

न्यूनतम9.2 tok/s

Eval score· 1 evalBF16

81.7%

Qwentext-generation14 स्पीड टेस्ट

Ornith-1.0-35B

ornith-ai / Ornith-1.0-35B

MoE35Bqwen3_5_moeimage-text-to-textconversational

Qwen3-8B-Base

Qwen / Qwen3-8B-Base

8Bqwen3conversational

Qwen3.5-4B-Base

Qwen / Qwen3.5-4B-Base

4Bqwen3_5conversational

Qwenimage-text-to-text19 स्पीड टेस्ट

Qwen3.5-27B

Qwen / Qwen3.5-27B

27Bqwen3_5conversational

LFM2.5-8B-A1B-Base

LiquidAI / LFM2.5-8B-A1B-Base

MoE1B active / 8B totallfm2_moeliquidlfm2.5

GLM-5.2

zai-org / GLM-5.2

MoE753Bglm_moe_dsaconversationalen

Deepseektext-generation38 स्पीड टेस्ट

DeepSeek-V4-Flash

deepseek-ai / DeepSeek-V4-Flash

MoE291Bdeepseek_v48-bitfp8

सर्वश्रेष्ठ262 tok/s

मीडियन28.5 tok/s

न्यूनतम7.0 tok/s

Eval score· 1 evalIQ3_XXS

94.8%

text-generation47 स्पीड टेस्ट

Laguna-S-2.1

poolside / Laguna-S-2.1

MoE118Blagunalaguna-s-2.1vllm

Qwen3-Coder-Next

Qwen / Qwen3-Coder-Next

Qwentext-generation80 स्पीड टेस्ट

MoE80Bqwen3_nextconversational

Muse-Glimmer-30B

meta-models / Muse-Glimmer-30B

image-text-to-text49 स्पीड टेस्ट

30Bmuse_glimmerconversational

सर्वश्रेष्ठ1.5k tok/s

मीडियन68.4 tok/s

न्यूनतम12.8 tok/s

Qwentext-generation2 स्पीड टेस्ट

Qwen2.5-7B

Qwen / Qwen2.5-7B

7Bqwen2conversationalen

सर्वश्रेष्ठ1.4k tok/s

मीडियन1.4k tok/s

न्यूनतम1.4k tok/s

0.3Bintern_vit_6bfeature-extraction

InternViT-300M-448px

OpenGVLab / InternViT-300M-448px

image-feature-extraction

Phitext-generation17 स्पीड टेस्ट

Phi-4-mini-reasoning

microsoft / Phi-4-mini-reasoning

4Bphi3nlpmath

सर्वश्रेष्ठ6.1k tok/s

मीडियन130 tok/s

न्यूनतम32.9 tok/s

Qwen3-0.6B-Base

Qwen / Qwen3-0.6B-Base

0.6Bqwen3conversational

Gpttext-generation56 स्पीड टेस्ट

gpt-oss-120b

openai / gpt-oss-120b

MoE120Bgpt_ossvllmconversational

सर्वश्रेष्ठ2.8k tok/s

मीडियन115 tok/s

न्यूनतम8.8 tok/s

Eval score· 2 evalsMXFP4

78.4%

Llamatext-generation50 स्पीड टेस्ट

Llama-3.2-3B-Instruct

meta-llama / Llama-3.2-3B-Instruct

3Bfacebookmetallama-3

Ornith-1.5-9B-GGUF

ornith-ai / Ornith-1.5-9B-GGUF

text-generation54 स्पीड टेस्ट

9Bconversational

सर्वश्रेष्ठ92.0 tok/s

मीडियन41.6 tok/s

न्यूनतम14.0 tok/s

Qwentext-generation29 स्पीड टेस्ट

Qwen3-Coder-30B-A3B-Instruct

Qwen / Qwen3-Coder-30B-A3B-Instruct

सर्वश्रेष्ठ136 tok/s

मीडियन80.5 tok/s

न्यूनतम14.4 tok/s

Eval score· 1 evalQ4_K_M

92.1%

LFM2.5-1.2B-Base

LiquidAI / LFM2.5-1.2B-Base

1.2Blfm2liquidlfm2.5

text-generation39 स्पीड टेस्ट

LFM2-350M

LiquidAI / LFM2-350M

gemma-4-E2B

google / gemma-4-E2B

Gemmaany-to-any

2Bgemma4image-text-to-text

Qwentext-generation44 स्पीड टेस्ट

Ornith-1.0-9B

ornith-ai / Ornith-1.0-9B

9Bqwen3_5image-text-to-textconversational

सर्वश्रेष्ठ77.5 tok/s

मीडियन25.7 tok/s

न्यूनतम3.8 tok/s

Eval score· 3 evalsQ4_K_M

56.0%

text-generation21 स्पीड टेस्ट

GLM-4.7-Flash

zai-org / GLM-4.7-Flash

MoE31Bglm4_moe_liteconversationalen

Qwen3-14B-Base

Qwen / Qwen3-14B-Base

14Bqwen3conversational

text-generation6 स्पीड टेस्ट

Laguna-XS-2.1

poolside / Laguna-XS-2.1

MoE33Blagunalaguna-xs-2.1vllm

Llama-4-Scout-17B-16E

meta-llama / Llama-4-Scout-17B-16E

Llamaimage-text-to-text

MoE17B active / 109B totalllama4facebookmeta

gemma-3-12b-pt

google / gemma-3-12b-pt

12Bgemma3

Nanbeige4.3-33B-Base

Nanbeige / Nanbeige4.3-33B-Base

33Bllmnanbeigeconversational

Llamatext-generation40 स्पीड टेस्ट

MiniCPM5-1B-GGUF

openbmb / MiniCPM5-1B-GGUF

1Bminicpmminicpm5long-context

Qwen3-1.7B-Base

Qwen / Qwen3-1.7B-Base

1.7Bqwen3conversational

Llama-3.1-70B

meta-llama / Llama-3.1-70B

70Bfacebookmetallama-3

Ministral-3-3B-Base-2512

mistralai / Ministral-3-3B-Base-2512

3Bvllmmistral3mistral-common

Deepseek-R1text-generation28 स्पीड टेस्ट

DeepSeek-R1-Distill-Qwen-7B

deepseek-ai / DeepSeek-R1-Distill-Qwen-7B

7Bqwen2conversational

Ornstein3.6-27B-MTP-NSC-ACE-SABER

GestaltLabs / Ornstein3.6-27B-MTP-NSC-ACE-SABER

27Bqwen3_5qwen3.6mtp

Qwen2.5-72B

Qwen / Qwen2.5-72B

72Bqwen2conversationalen

Qwen2.5-32B

Qwen / Qwen2.5-32B

32Bqwen2conversationalen

Qwentext-generation28 स्पीड टेस्ट

AceReason-Nemotron-1.1-7B

nvidia / AceReason-Nemotron-1.1-7B

7Bqwen2nvidiareasoning

सर्वश्रेष्ठ79.8 tok/s

मीडियन63.6 tok/s

न्यूनतम12.8 tok/s

text-generation24 स्पीड टेस्ट

LFM2-8B-A1B

LiquidAI / LFM2-8B-A1B

MoE1B active / 8B totallfm2_moeliquidlfm2

phi-4

microsoft / phi-4

Phitext-generation3 स्पीड टेस्ट

15Bphi3nlpmath

सर्वश्रेष्ठ77.0 tok/s

मीडियन36.0 tok/s

न्यूनतम36.0 tok/s

Qwentext-generation9 स्पीड टेस्ट

Nex-N2-mini

nex-agi / Nex-N2-mini

MoE35Bqwen3_5_moeimage-text-to-textconversational

Mistral-Small-24B-Base-2501

mistralai / Mistral-Small-24B-Base-2501

24Bvllmmistral-commonen

Deepseek-R1text-generation10 स्पीड टेस्ट

DeepSeek-R1-Distill-Qwen-14B

deepseek-ai / DeepSeek-R1-Distill-Qwen-14B

14Bqwen2conversational

सर्वश्रेष्ठ50.9 tok/s

मीडियन38.0 tok/s

न्यूनतम9.0 tok/s

text-generation33 स्पीड टेस्ट

Ornith-1.0-9B-GGUF

ornith-ai / Ornith-1.0-9B-GGUF

9Bconversational

सर्वश्रेष्ठ88.1 tok/s

LFM2.5-350M-Base

LiquidAI / LFM2.5-350M-Base

350Mlfm2liquidlfm2.5

Deepseekimage-text-to-text30 स्पीड टेस्ट

DeepSeek-V4.1-Flash

deepseek-ai / DeepSeek-V4.1-Flash

763Bdeepseek_v41text-generationconversational

Nemotron-Cascade-2-30B-A3B

nvidia / Nemotron-Cascade-2-30B-A3B

text-generation16 स्पीड टेस्ट

MoE3B active / 30B totalnemotron_hnvidianemotron-cascade-2

Meta-Llama-3.1-70B-Instruct-AWQ-INT4

hugging-quants / Meta-Llama-3.1-70B-Instruct-AWQ-INT4

Llamatext-generation31 स्पीड टेस्ट

70Bllama-3.1metaautoawq

Meta-Llama-3-8B-Instruct

meta-llama / Meta-Llama-3-8B-Instruct

Llamatext-generation31 स्पीड टेस्ट

8Bfacebookmetallama-3

Ornith-1.0-35B-GGUF

ornith-ai / Ornith-1.0-35B-GGUF

text-generation31 स्पीड टेस्ट

Qwen3-4B-Base

Qwen / Qwen3-4B-Base

4Bqwen3conversational

Qwen2.5-3B

Qwen / Qwen2.5-3B

3Bqwen2conversationalen

GLM-5.3

zai-org / GLM-5.3

MoE753Bglm_moe_dsaconversationalen

image-text-to-text21 स्पीड टेस्ट

NVIDIA-Nemotron-3.5-Super-VL-120B-A12B-BF16

nvidia / NVIDIA-Nemotron-3.5-Super-VL-120B-A12B-BF16

MoE12B active / 120B totalnemotron_h_omninvidianemotron-3.5

सर्वश्रेष्ठ33.6 tok/s

मीडियन20.6 tok/s

न्यूनतम5.3 tok/s

Qwentext-generation12 स्पीड टेस्ट

Agents-A1

InternScience / Agents-A1

MoE35Bqwen3_5_moeimage-text-to-textmoe

सर्वश्रेष्ठ96.4 tok/s

Qwen3-30B-A3B-Instruct-2507

Qwen / Qwen3-30B-A3B-Instruct-2507

Qwenimage-text-to-text1 स्पीड टेस्ट

Qwen3.5-0.8B-Base

Qwen / Qwen3.5-0.8B-Base

0.8Bqwen3_5conversational

सर्वश्रेष्ठ2.7k tok/s

मीडियन2.7k tok/s

न्यूनतम2.7k tok/s

LFM2.5-230M-Base

LiquidAI / LFM2.5-230M-Base

230Mlfm2liquidlfm2.5

Trinity-Mini-Base-Pre-Anneal

arcee-ai / Trinity-Mini-Base-Pre-Anneal

Arceetext-generation

MoE26Bafmoeenes

MoE3B active / 30B totalconversational

NVIDIA-Nemotron-3.5-Lightning-30B-A3B-GGUF

lmstudio-community / NVIDIA-Nemotron-3.5-Lightning-30B-A3B-GGUF

26 स्पीड टेस्ट

GLM-4.5-Air

zai-org / GLM-4.5-Air

text-generation9 स्पीड टेस्ट

MoE110Bglm4_moeconversationalen

सर्वश्रेष्ठ2.5k tok/s

मीडियन103 tok/s

न्यूनतम6.2 tok/s

Qwentext-generation5 स्पीड टेस्ट

Qwen3-32B

Qwen / Qwen3-32B

32Bqwen3conversational

सर्वश्रेष्ठ79.3 tok/s

मीडियन22.8 tok/s

न्यूनतम3.2 tok/s

Qwentext-generation25 स्पीड टेस्ट

Qwen3-Coder-30B-A3B-Instruct-FP8

Qwen / Qwen3-Coder-30B-A3B-Instruct-FP8

MoE3B active / 30B totalqwen3_moeconversationalfp8

सर्वश्रेष्ठ3.1k tok/s

मीडियन871 tok/s

न्यूनतम26.1 tok/s

Qwen2.5-14B

Qwen / Qwen2.5-14B

14Bqwen2conversationalen

Llamatext-generation23 स्पीड टेस्ट

MiniCPM5-1B

openbmb / MiniCPM5-1B

1Bminicpmminicpm5long-context

सर्वश्रेष्ठ23.1k tok/s

Ornith-1.5-122B

ornith-ai / Ornith-1.5-122B

Qwentext-generation23 स्पीड टेस्ट

MoE122Bqwen3_5_moeimage-text-to-textconversational

Step-3.5-Flash

stepfun-ai / Step-3.5-Flash

text-generation23 स्पीड टेस्ट

199Bstep3p5conversational

सर्वश्रेष्ठ31.6 tok/s

मीडियन21.5 tok/s

न्यूनतम7.5 tok/s

reka-flash-3

RekaAI / reka-flash-3

Llama23 स्पीड टेस्ट

21B

सर्वश्रेष्ठ36.3 tok/s

मीडियन23.1 tok/s

न्यूनतम6.0 tok/s

text-generation22 स्पीड टेस्ट

MiMo-V2.6-Flash-RL

XiaomiMiMo / MiMo-V2.6-Flash-RL

MoE311Bmimo_v2multimodalvision-language

GLM-5.3-Flash-BF16

zai-org / GLM-5.3-Flash-BF16

321Bglm5_nextconversationalen

BTL-4

badtheorylabs / BTL-4

MoE35Bqwen3_5_moeimage-text-to-textagentic

7Bhunyuan_v1_densetext-generationzh

Hy-MT2-7B

tencent / Hy-MT2-7B

translation7 स्पीड टेस्ट

सर्वश्रेष्ठ98.1 tok/s

NVIDIA-Nemotron-3.5-Lightning-30B-A3B-NVFP4

nvidia / NVIDIA-Nemotron-3.5-Lightning-30B-A3B-NVFP4

Opttext-generation20 स्पीड टेस्ट

MoE3B active / 30B totalnemotron_hnvidianemotron-3.5

सर्वश्रेष्ठ6.7k tok/s

मीडियन293 tok/s

न्यूनतम77.1 tok/s

Eval score· 3 evalsNVFP4

74.8%

Deepseek-R1text-generation4 स्पीड टेस्ट

DeepSeek-R1-Distill-Qwen-32B

deepseek-ai / DeepSeek-R1-Distill-Qwen-32B

32Bqwen2conversational

सर्वश्रेष्ठ23.3 tok/s

मीडियन22.3 tok/s

न्यूनतम21.4 tok/s

text-generation16 स्पीड टेस्ट

Ling-3.0-tiny

inclusionAI / Ling-3.0-tiny

MoE8Bbailing_hybridconversational

सर्वश्रेष्ठ3.1k tok/s

मीडियन143 tok/s

न्यूनतम52.6 tok/s

Qwentext-generation18 स्पीड टेस्ट

Nemotron-Cascade-14B-Thinking

nvidia / Nemotron-Cascade-14B-Thinking

14Bqwen3nvidianemotron-cascade

सर्वश्रेष्ठ51.9 tok/s

मीडियन34.2 tok/s

न्यूनतम6.2 tok/s

Olmo-3-1125-32B

allenai / Olmo-3-1125-32B

32Bolmo3en

text-generation17 स्पीड टेस्ट

GLM-Z1-32B-0414

zai-org / GLM-Z1-32B-0414

32Bglm4conversationalzh

सर्वश्रेष्ठ22.5 tok/s

मीडियन16.4 tok/s

न्यूनतम5.7 tok/s

Mellum2-12B-A2.5B-Instruct

JetBrains / Mellum2-12B-A2.5B-Instruct

MoE2.5B active / 12B totalmellumconversationalen

सर्वश्रेष्ठ83.5 tok/s

मीडियन51.8 tok/s

न्यूनतम41.8 tok/s

Coheretext-generation17 स्पीड टेस्ट

North-Mini-Code-1.0

CohereLabs / North-Mini-Code-1.0

MoE30Bcohere2_moeconversationalchat

Qwen3.5-122B-A10B-GPTQ-Int4

Qwen / Qwen3.5-122B-A10B-GPTQ-Int4

Qwenimage-text-to-text17 स्पीड टेस्ट

MoE10B active / 122B totalqwen3_5_moeconversational4-bit

सर्वश्रेष्ठ1.8k tok/s

मीडियन540 tok/s

न्यूनतम49.1 tok/s

image-text-to-text16 स्पीड टेस्ट

GLM-4.6V

zai-org / GLM-4.6V

MoE108Bglm4v_moeconversationalzh

सर्वश्रेष्ठ29.6 tok/s

मीडियन16.9 tok/s

न्यूनतम5.7 tok/s

Qwen3.5-2B-Base

Qwen / Qwen3.5-2B-Base

2Bqwen3_5conversational

Llama-3.1-8B

meta-llama / Llama-3.1-8B

8Bfacebookmetallama-3

image-text-to-text5 स्पीड टेस्ट

Step-3.7-Flash

stepfun-ai / Step-3.7-Flash

MoE201Bstep3p7vision-languagemultimodal

सर्वश्रेष्ठ31.0 tok/s

मीडियन30.8 tok/s

न्यूनतम30.7 tok/s

gemma-3-1b-pt

google / gemma-3-1b-pt

Gemmatext-generation

1Bgemma3_text

MoE3B active / 30B totalNemotronH_Nano_Omni_Reasoning_V3feature-extractionnvidia

Nemotron-3-Nano-Omni-30B-A3B-Reasoning-BF16

nvidia / Nemotron-3-Nano-Omni-30B-A3B-Reasoning-BF16

any-to-any5 स्पीड टेस्ट

Agents-A1-4B

InternScience / Agents-A1-4B

4Bqwen3_5image-text-to-textvlm

QUEST-30B-RL

osunlp / QUEST-30B-RL

MoE30Bqwen3_moequestconversational

6Bintern_vit_6bfeature-extraction

InternViT-6B-224px

OpenGVLab / InternViT-6B-224px

image-feature-extraction

Qwen3-30B-A3B-Base

Qwen / Qwen3-30B-A3B-Base

KAT-Coder-V2.5-Dev

Kwaipilot / KAT-Coder-V2.5-Dev

MoE35Bqwen3_5_moeimage-text-to-textcode

सर्वश्रेष्ठ75.9 tok/s

मीडियन75.0 tok/s

न्यूनतम66.3 tok/s

Mistral-Medium-3.5-128B

mistralai / Mistral-Medium-3.5-128B

Mistral13 स्पीड टेस्ट

LFM2.5-2.6B-Base

LiquidAI / LFM2.5-2.6B-Base

2.6Blfm2liquidlfm2.5

MoE3B active / 35B totalqwen3_5_moe

Tess-4-35B-A3B

migtissera / Tess-4-35B-A3B

Qwen12 स्पीड टेस्ट

granite-4.0-h-micro

ibm-granite / granite-4.0-h-micro

text-generation12 स्पीड टेस्ट

MoE3Bgranitemoehybridlanguagegranite-4.0

सर्वश्रेष्ठ141 tok/s

मीडियन41.9 tok/s

न्यूनतम7.9 tok/s

Eval score· 3 evalsQ4_K_M

53.3%

SmolLM2-135M

HuggingFaceTB / SmolLM2-135M

135Men

text-generation11 स्पीड टेस्ट

GLM-Z1-Rumination-32B-0414

zai-org / GLM-Z1-Rumination-32B-0414

32Bglm4conversationalzh

सर्वश्रेष्ठ22.4 tok/s

मीडियन19.7 tok/s

न्यूनतम8.8 tok/s

Minimaxtext-generation2 स्पीड टेस्ट

MiniMax-M2.5

MiniMaxAI / MiniMax-M2.5

Llamatext-generation7 स्पीड टेस्ट

Llama-3.2-1B-Instruct

meta-llama / Llama-3.2-1B-Instruct

1Bfacebookmetallama-3

SmolLM2-360M

HuggingFaceTB / SmolLM2-360M

360Men

Mistral-Small-3.1-24B-Base-2503

mistralai / Mistral-Small-3.1-24B-Base-2503

24Bvllmmistral3mistral-common

Seed-OSS-36B-Instruct

ByteDance-Seed / Seed-OSS-36B-Instruct

36Bseed_ossvllmconversational

Qwentext-generation10 स्पीड टेस्ट

Qwen3.6-27B-DFlash

z-lab / Qwen3.6-27B-DFlash

27Bqwen3feature-extractiondflash

Spark-X2.5-4B-Base

XHToken / Spark-X2.5-4B-Base

4Bspark2_5llmsparkx2_5

NVIDIA-Nemotron-3-Nano-30B-A3B-BF16

nvidia / NVIDIA-Nemotron-3-Nano-30B-A3B-BF16

MoE3B active / 30B totalnemotron_hnvidiaconversational

Ling-3.0-flash

inclusionAI / Ling-3.0-flash

text-generation9 स्पीड टेस्ट

MoE127Bbailing_hybridconversational

सर्वश्रेष्ठ72.3 tok/s

मीडियन43.4 tok/s

न्यूनतम33.6 tok/s

7Bhunyuan_v1_densetext-generation

Hunyuan-MT-7B

tencent / Hunyuan-MT-7B

translation9 स्पीड टेस्ट

सर्वश्रेष्ठ98.1 tok/s

मीडियन74.3 tok/s

न्यूनतम28.5 tok/s

text-generation5 स्पीड टेस्ट

MiMo-V2.5

XiaomiMiMo / MiMo-V2.5

MoE311Bmimo_v2multimodalvision-language

सर्वश्रेष्ठ33.9 tok/s

मीडियन32.7 tok/s

न्यूनतम28.1 tok/s

Qwen2.5-0.5B

Qwen / Qwen2.5-0.5B

0.5Bqwen2conversationalen

MoE3B active / 30B totalhy_v3text-generationzh

Hy-MT2-30B-A3B

tencent / Hy-MT2-30B-A3B

translation

8Binternlm3conversational

internlm3-8b-instruct

internlm / internlm3-8b-instruct

Internlmtext-generation

Nemotron-Cascade-8B

nvidia / Nemotron-Cascade-8B

8Bqwen3nvidiaNemotron-Cascade

UI-TARS-7B-DPO

ByteDance-Seed / UI-TARS-7B-DPO

7Bqwen2_vlmultimodalgui

text-generation8 स्पीड टेस्ट

EXAONE-3.5-7.8B-Instruct

LGAI-EXAONE / EXAONE-3.5-7.8B-Instruct

7.8Bexaonelg-aiexaone-3.5

सर्वश्रेष्ठ26.8 tok/s

मीडियन25.7 tok/s

न्यूनतम11.2 tok/s

text-generation8 स्पीड टेस्ट

Ling-2.6-flash

inclusionAI / Ling-2.6-flash

MoE107Bbailing_hybridconversationalen

सर्वश्रेष्ठ94.9 tok/s

मीडियन86.2 tok/s

न्यूनतम82.3 tok/s

Gpttext-generation5 स्पीड टेस्ट

Tinystories-gpt-0.1-3m

segestic / Tinystories-gpt-0.1-3m

3Mgpt2en

सर्वश्रेष्ठ5.0k tok/s

मीडियन4.9k tok/s

न्यूनतम1.3k tok/s

text-generation7 स्पीड टेस्ट

Kimi-Linear-48B-A3B-Instruct

moonshotai / Kimi-Linear-48B-A3B-Instruct

MoE3B active / 48B totalkimi_linearconversational

rwkv-7-world

BlinkDL / rwkv-7-world

Starcodertext-generation

rwkvcausal-lmen

granite-4.1-8b-base

ibm-granite / granite-4.1-8b-base

8Bgranitelanguagegranite-4.1

Sage-A1.5

InternScience / Sage-A1.5

MoE36Bqwen3_5_moeimage-text-to-textmoe

NVIDIA-Nemotron-3-Super-120B-A12B-BF16

nvidia / NVIDIA-Nemotron-3-Super-120B-A12B-BF16

MoE12B active / 120B totalnemotron_hnvidianemotron-3

Hy3

tencent / Hy3

MoE299Bhy_v3hunyuanhy3

Deepseek-Codertext-generation6 स्पीड टेस्ट

DeepSeek-Coder-V2-Lite-Instruct

deepseek-ai / DeepSeek-Coder-V2-Lite-Instruct

MoE16Bdeepseek_v2conversational

Kimi-K2.6

moonshotai / Kimi-K2.6

1Tkimi_k25feature-extractioncompressed-tensors

Llamatext-generation4 स्पीड टेस्ट

MiniCPM5-2B

openbmb / MiniCPM5-2B

2Bminicpmminicpm5long-context

glm-4-9b-chat

zai-org / glm-4-9b-chat

Chatglm

9Bglmthudmzh

Qwentext-generation5 स्पीड टेस्ट

Qwen3-Coder-Next-FP8

Qwen / Qwen3-Coder-Next-FP8

MoE80Bqwen3_nextconversationalfp8

Llama-2-7b-hf

meta-llama / Llama-2-7b-hf

Llamatext-generation5 स्पीड टेस्ट

7Bfacebookmetallama-2

Qwen3-VL-8B-Instruct

Qwen / Qwen3-VL-8B-Instruct

Qwenimage-text-to-text3 स्पीड टेस्ट

8Bqwen3_vlconversational

सर्वश्रेष्ठ95.9 tok/s

मीडियन89.8 tok/s

न्यूनतम58.8 tok/s

Qwen2.5-1.5B

Qwen / Qwen2.5-1.5B

1.5Bqwen2conversationalen

Opttext-generation5 स्पीड टेस्ट

NVIDIA-Nemotron-3-Super-120B-A12B-NVFP4

nvidia / NVIDIA-Nemotron-3-Super-120B-A12B-NVFP4

MoE12B active / 120B totalnemotron_hnvidianemotron-3

Qwen3.5-397B-A17B

Qwen / Qwen3.5-397B-A17B

MoE17B active / 397B totalqwen3_5_moeconversational

gemma-3n-E4B

google / gemma-3n-E4B

4Bgemma3nautomatic-speech-recognitionautomatic-speech-translation

Spark-X2.6-4B-Base

XHToken / Spark-X2.6-4B-Base

4Bspark2_5llmsparkx2_5

Qwen3-235B-A22B

Qwen / Qwen3-235B-A22B

MoE22B active / 235B totalqwen3_moeconversational

Spark-X2.5-1.7B-Base

XHToken / Spark-X2.5-1.7B-Base

1.7Bspark2_5llmsparkx2_5

Qwen3-235B-A22B-Thinking-2507

Qwen / Qwen3-235B-A22B-Thinking-2507

MoE22B active / 235B totalqwen3_moeconversational

सर्वश्रेष्ठ17.5 tok/s

मीडियन15.8 tok/s

न्यूनतम13.0 tok/s

Qwen2.5-VL-7B-Instruct

Qwen / Qwen2.5-VL-7B-Instruct

7Bqwen2_5_vlmultimodalconversational

NVIDIA-Nemotron-3-Nano-4B-FP8

nvidia / NVIDIA-Nemotron-3-Nano-4B-FP8

4Bnemotron_hnvidiaconversational

Deepseektext-generation4 स्पीड टेस्ट

DeepSeek-V4-Flash-DSpark

deepseek-ai / DeepSeek-V4-Flash-DSpark

MoE165Bdeepseek_v48-bitfp8

gemma-3-4b-pt

google / gemma-3-4b-pt

4Bgemma3

Minimaxtext-generation2 स्पीड टेस्ट

MiniMax-M2

MiniMaxAI / MiniMax-M2

GLM-5.1

zai-org / GLM-5.1

MoE754Bglm_moe_dsaconversationalen

text-generation4 स्पीड टेस्ट

granite-4.1-30b

ibm-granite / granite-4.1-30b

30Bgranitelanguagegranite-4.1

सर्वश्रेष्ठ17.9 tok/s

मीडियन16.4 tok/s

न्यूनतम15.6 tok/s

Gemmatext-generation4 स्पीड टेस्ट

Gemopus-4-26B-A4B-it

Jackrong / Gemopus-4-26B-A4B-it

MoE4B active / 26B totalgemma4instruction-tunedreasoning

सर्वश्रेष्ठ64.3 tok/s

मीडियन55.0 tok/s

न्यूनतम45.7 tok/s

Llamatext-generation3 स्पीड टेस्ट

MiniCPM5-2B-GGUF

openbmb / MiniCPM5-2B-GGUF

2Bminicpmminicpm5long-context

सर्वश्रेष्ठ60.5 tok/s

मीडियन39.1 tok/s

न्यूनतम33.7 tok/s

gemma-3-27b-pt

google / gemma-3-27b-pt

27Bgemma3

Nex-N2.5-mini

nex-agi / Nex-N2.5-mini

MoE35Bqwen3_5_moeimage-text-to-textconversational

Qwen3-VL-30B-A3B-Instruct

Qwen / Qwen3-VL-30B-A3B-Instruct

MoE3B active / 30B totalqwen3_vl_moeconversational

सर्वश्रेष्ठ56.6 tok/s

मीडियन52.2 tok/s

न्यूनतम47.7 tok/s

Falcon-H1-0.5B-Base

tiiuae / Falcon-H1-0.5B-Base

Falcontext-generation

0.5Bfalcon_h1en

Gemmatext-generation3 स्पीड टेस्ट

functiongemma-270m-it

google / functiongemma-270m-it

270Mgemma3_textgemma3google

Falcon-H1-Tiny-90M-Instruct

tiiuae / Falcon-H1-Tiny-90M-Instruct

Falcontext-generation3 स्पीड टेस्ट

90Mfalcon_h1edgeconversational

gemma-3-270m

google / gemma-3-270m

Gemmatext-generation

270Mgemma3_textgemma3google

EXAONE-3.5-7.8B-Instruct-GGUF

LGAI-EXAONE / EXAONE-3.5-7.8B-Instruct-GGUF

7.8Blg-aiexaoneexaone-3.5

सर्वश्रेष्ठ27.2 tok/s

Ornith-1.0-397B

ornith-ai / Ornith-1.0-397B

MoE397Bqwen3_5_moeimage-text-to-textconversational

Mellum2-12B-A2.5B-Thinking

JetBrains / Mellum2-12B-A2.5B-Thinking

MoE2.5B active / 12B totalmellumconversationalen

Gemmaimage-text-to-text1 स्पीड टेस्ट

diffusiongemma-26B-A4B-it

google / diffusiongemma-26B-A4B-it

MoE4B active / 26B totaldiffusion_gemmaconversational

MiniCPM3-4B

openbmb / MiniCPM3-4B

4Bminicpm3conversationalzh

सर्वश्रेष्ठ35.6 tok/s

NVIDIA-Nemotron-Nano-9B-v2-quantized.w4a16

RedHatAI / NVIDIA-Nemotron-Nano-9B-v2-quantized.w4a16

9Bnemotron_hnvidiaint4

LFM3-1.7B-Base

LiquidAI / LFM3-1.7B-Base

1.7Blfm3liquidedge

Groot-19B-A3B-261023-NVFP4

GrootTeam / Groot-19B-A3B-261023-NVFP4

MoE3B active / 19B totalqwen3_5_moeconversational8-bit

सर्वश्रेष्ठ426 tok/s

मीडियन416 tok/s

न्यूनतम406 tok/s

Eval score· 2 evalsNVFP4

84.0%

image-text-to-text1 स्पीड टेस्ट

Kimi-K2.5

moonshotai / Kimi-K2.5

1Tkimi_k25feature-extractioncompressed-tensors

सर्वश्रेष्ठ74.0 tok/s

मीडियन74.0 tok/s

न्यूनतम74.0 tok/s

Mistraltext-generation2 स्पीड टेस्ट

Mistral-7B-Instruct-v0.3-GGUF

bartowski / Mistral-7B-Instruct-v0.3-GGUF

7Bconversational

सर्वश्रेष्ठ53.2 tok/s

मीडियन50.2 tok/s

न्यूनतम47.1 tok/s

flint

RecursiveSI / flint

MoE40Bflintconversational

Minimaximage-text-to-text2 स्पीड टेस्ट

MiniMax-M3.1-Flash

MiniMaxAI / MiniMax-M3.1-Flash

MoE220Bminimax_m3_vlmultimodalmoe

Qwen3-235B-A22B-Instruct-2507

Qwen / Qwen3-235B-A22B-Instruct-2507

Qwentext-generation2 स्पीड टेस्ट

MoE22B active / 235B totalqwen3_moeconversational

सर्वश्रेष्ठ72.9 tok/s

मीडियन72.6 tok/s

न्यूनतम72.3 tok/s

SmolLM2-1.7B

HuggingFaceTB / SmolLM2-1.7B

1.7Ben

Qwen3-4B-Thinking-2507

Qwen / Qwen3-4B-Thinking-2507

4Bqwen3conversational

TinyLlama-1.1B-Chat-v1.0

TinyLlama / TinyLlama-1.1B-Chat-v1.0

Ternary-Bonsai-8B-unpacked

prism-ml / Ternary-Bonsai-8B-unpacked

8Bqwen3prismmlbonsai

Deepseekimage-text-to-text1 स्पीड टेस्ट

DeepSeek-V4-Flash-Vision-Exp

deepseek-ai / DeepSeek-V4-Flash-Vision-Exp

MoE305Bdeepseek_v4text-generation8-bit

Qwen3.8-27B-Uncensored-int4-AutoRound

johannrplaster / Qwen3.8-27B-Uncensored-int4-AutoRound

27Bqwen3_5vllmintel-xpu

stories-llama2-50k

delphi-suite / stories-llama2-50k

Llama2 स्पीड टेस्ट

delphien

सर्वश्रेष्ठ306.1k tok/s

मीडियन171.4k tok/s

न्यूनतम36.7k tok/s

LFM2-24B-A2B

LiquidAI / LFM2-24B-A2B

MoE2B active / 24B totallfm2_moeliquidlfm2

models-moved

ggml-org / models-moved

2 स्पीड टेस्ट

सर्वश्रेष्ठ16.7k tok/s

मीडियन10.6k tok/s

न्यूनतम4.4k tok/s

Qwentext-generation2 स्पीड टेस्ट

Qwen3.6-27b-ROCmFP4-CIRU-CHAOS-SPEED

jcbtc / Qwen3.6-27b-ROCmFP4-CIRU-CHAOS-SPEED

Eval score· 1 evalROCmFP4

90.5%

MobileLLM-350M

facebook / MobileLLM-350M

350Mmobilellmfacebookmeta

MobileLLM-125M

facebook / MobileLLM-125M

125Mmobilellmfacebookmeta

सर्वश्रेष्ठ1.3k tok/s

मीडियन987 tok/s

न्यूनतम703 tok/s

OLMo-2-0425-1B

allenai / OLMo-2-0425-1B

1Bolmo2en

LFM2-700M

LiquidAI / LFM2-700M

Llamatext-generation2 स्पीड टेस्ट

h2o-danube3-500m-chat

h2oai / h2o-danube3-500m-chat

0.5Bgptllmlarge language model

granite-4.0-1b-base

ibm-granite / granite-4.0-1b-base

MoE1Bgranitemoehybridlanguagegranite-4.0

Phi-4-mini-instruct

microsoft / Phi-4-mini-instruct

Phitext-generation

4Bphi3nlpcode

Minimaxtext-generation2 स्पीड टेस्ट

MiniMax-M2.1

MiniMaxAI / MiniMax-M2.1

Qwen3.5-35B-A3B-4bit

mlx-community / Qwen3.5-35B-A3B-4bit

MoE3B active / 35B totalqwen3_5_moemlxconversational

MiniMax-M3

MiniMaxAI / MiniMax-M3

Minimaximage-text-to-text

MoE427Bminimax_m3_vlmultimodalmoe

Qwen3-4B-Instruct-2507

Qwen / Qwen3-4B-Instruct-2507

4Bqwen3conversational

VertaLily-1.2-1B-GGUF

VLTX / VertaLily-1.2-1B-GGUF

Optimage-text-to-text1 स्पीड टेस्ट

1Bvltxsovereigntechina-x

gpt-5.6-sol

oroboros-labs / gpt-5.6-sol

Comet_60B_A12B_NVFP4

MultiverseComputingCAI / Comet_60B_A12B_NVFP4

60Bnemotron_h_omniconversationalen

सर्वश्रेष्ठ18.2 tok/s

मीडियन18.2 tok/s

न्यूनतम18.2 tok/s

granite-4.0-350m-base

ibm-granite / granite-4.0-350m-base

MoE350Mgranitemoehybridlanguagegranite-4.0

LaMini-GPT-774M

MBZUAI / LaMini-GPT-774M

Gpttext-generation

774Mgpt2en

granite-docling-258M

ibm-granite / granite-docling-258M

258Midefics3text-generationdocuments

Qwentext-to-speech1 स्पीड टेस्ट

bge-base-en-v1.5-gguf

CompendiumLabs / bge-base-en-v1.5-gguf

VieNeu-TTS-0.3B

pnnbao-ump / VieNeu-TTS-0.3B

TIPO-100M

KBlueLeaf / TIPO-100M

0.1Ben

सर्वश्रेष्ठ1.7k tok/s

मीडियन1.7k tok/s

न्यूनतम1.7k tok/s

SmolLM2-135M-GGUF

QuantFactory / SmolLM2-135M-GGUF

harrier-oss-v1-0.6b

calculatortamer / harrier-oss-v1-0.6b

0.6Bsentence-transformersqwen3mteb

Phitext-to-speech1 स्पीड टेस्ट

VieNeu-TTS-v2-Turbo-GGUF

pnnbao-ump / VieNeu-TTS-v2-Turbo-GGUF

ttsvietnamesebilingual

सर्वश्रेष्ठ1.2k tok/s

मीडियन1.2k tok/s

न्यूनतम1.2k tok/s

test-gguf

rarad / test-gguf

TinyStories-656K

raincandy-u / TinyStories-656K

llama2tinyllamasvalidation

tiny1m

shibatch / tiny1m

Llama1 स्पीड टेस्ट

सर्वश्रेष्ठ5.0k tok/s

मीडियन5.0k tok/s

न्यूनतम5.0k tok/s

SmolLM-135M

HuggingFaceTB / SmolLM-135M

135Men

MiniLM-L12-H384-uncased

microsoft / MiniLM-L12-H384-uncased

text-classification

bert

30Msentence-transformersbertsentence-similarity

bge-small-en-v1.5

BAAI / bge-small-en-v1.5

feature-extraction

1Bultravoxfeature-extractionar

SmolLM2-360M-GGUF

QuantFactory / SmolLM2-360M-GGUF

ultravox-v0_5-llama-3_2-1b

fixie-ai / ultravox-v0_5-llama-3_2-1b

Llamaaudio-text-to-text

Qwentext-generation1 स्पीड टेस्ट

Qwen3.5-2B-Claude-4.6-Opus-Reasoning-Distilled-GGUF

Jackrong / Qwen3.5-2B-Claude-4.6-Opus-Reasoning-Distilled-GGUF

2Bqwen3_5unslothqwen3.5-2B

Qwen2-0.5B

Qwen / Qwen2-0.5B

0.5Bqwen2pretrainedconversational

Hare-1.1B-base-0.5v-GGUF

mradermacher / Hare-1.1B-base-0.5v-GGUF

MiniCPM5-2B-LocalAI

weareapexcreators / MiniCPM5-2B-LocalAI

LFM2.5-230M-LocalAI

weareapexcreators / LFM2.5-230M-LocalAI

1 स्पीड टेस्ट

230Mconversational

सर्वश्रेष्ठ72.2 tok/s

मीडियन72.2 tok/s

न्यूनतम72.2 tok/s

Qwen3.5-4B-MTP-LocalAI

weareapexcreators / Qwen3.5-4B-MTP-LocalAI

0.6Bpii_maskingpiiprivacy

4Bimatrixconversational

PII-Tracer

perplexity-ai / PII-Tracer

Qwentoken-classification

Granite-4.1-3B-LocalAI

weareapexcreators / Granite-4.1-3B-LocalAI

1 स्पीड टेस्ट

3Bimatrixconversational

Qwen3.5-2B-MTP-LocalAI

weareapexcreators / Qwen3.5-2B-MTP-LocalAI

2Bimatrixconversational

Ornith-1.0-9B-MTP-LocalAI

weareapexcreators / Ornith-1.0-9B-MTP-LocalAI

LFM2.5-350M-LocalAI

weareapexcreators / LFM2.5-350M-LocalAI

Qwen3.5-0.8B-LocalAI

weareapexcreators / Qwen3.5-0.8B-LocalAI

0.8Bimatrixconversational

LFM2.5-1.2B-Instruct-LocalAI

weareapexcreators / LFM2.5-1.2B-Instruct-LocalAI

Gemma4-12B-Uncensored-LocalAI

weareapexcreators / Gemma4-12B-Uncensored-LocalAI

Gemma1 स्पीड टेस्ट

12Bconversational

सर्वश्रेष्ठ88.0 tok/s

मीडियन88.0 tok/s

न्यूनतम88.0 tok/s

Qwen3.5-4B-Uncensored-LocalAI

weareapexcreators / Qwen3.5-4B-Uncensored-LocalAI

TinyMistral-248M-v2

Locutusque / TinyMistral-248M-v2

Mistraltext-generation

248Men

Qwentext-to-speech1 स्पीड टेस्ट

VieNeu-TTS-v2

pnnbao-ump / VieNeu-TTS-v2

290Mqwen3podcastcode-switching

TinyLlama-1.1B-Chat-v0.3

TinyLlama / TinyLlama-1.1B-Chat-v0.3

1.1Ben

sentence-similarity1 स्पीड टेस्ट

sentence-bert-swedish-cased

KBLab / sentence-bert-swedish-cased

120Msentence-transformersbertfeature-extraction

Bonsai-27B-unpacked

prism-ml / Bonsai-27B-unpacked

27Bqwen3_5conversationalprismml

Bonsai-1.7B-unpacked

prism-ml / Bonsai-1.7B-unpacked

1.7Bqwen3prismmlbonsai

feature-extraction1 स्पीड टेस्ट

bge-small-en-v1.5-GGUF

unsloth / bge-small-en-v1.5-GGUF

sentence-transformerssentence-similaritymteb

Kolibri-1-BF16

Aleph-Alpha / Kolibri-1-BF16

MoE78Bkolibri1reasoningmoe

Ling-3.0-tiny-GGUF

inclusionAI / Ling-3.0-tiny-GGUF

qwen-3.8-27b-abliterated

heterodoxin / qwen-3.8-27b-abliterated

Ternary-Bonsai-1.7B-unpacked

prism-ml / Ternary-Bonsai-1.7B-unpacked

1.7Bqwen3prismmlbonsai

Ternary-Bonsai-2-27B-DFlash2-vllm

fraserprice / Ternary-Bonsai-2-27B-DFlash2-vllm

GLM-5.3-BF16

zai-org / GLM-5.3-BF16

MoE753Bglm_moe_dsaconversationalen

सर्वश्रेष्ठ6843.6k tok/s

मीडियन6843.6k tok/s

न्यूनतम6843.6k tok/s

Ling-3.0-flash-int4

inclusionAI / Ling-3.0-flash-int4

MoE127Bbailing_hybridconversationalcompressed-tensors

सर्वश्रेष्ठ37.3 tok/s

मीडियन37.3 tok/s

न्यूनतम37.3 tok/s

Falcon3-7B-Base

tiiuae / Falcon3-7B-Base

7Bfalcon3enfr

Qwentext-generation1 स्पीड टेस्ट

Ornith-1.5-35B-A3B-NVFP4

ornith-ai / Ornith-1.5-35B-A3B-NVFP4

MoE3B active / 35B totalqwen3_5_moeimage-text-to-textconversational

सर्वश्रेष्ठ259 tok/s

मीडियन259 tok/s

न्यूनतम259 tok/s

Eval score· 3 evalsNVFP4

70.9%

Qwen3-30B-A3B-Thinking-2507

Qwen / Qwen3-30B-A3B-Thinking-2507

bitnet-b1.58-2B-4T-bf16

microsoft / bitnet-b1.58-2B-4T-bf16

2Bbitnetchatlarge-language-model

Ministral-3-14B-Base-2512

mistralai / Ministral-3-14B-Base-2512

14Bvllmmistral3mistral-common

rnj-1-instruct-GGUF

EssentialAI / rnj-1-instruct-GGUF

GLM-4.6V-Flash

zai-org / GLM-4.6V-Flash

10Bglm4vconversationalzh

Deepseek-R1text-generation

gemma-3n-E4B-it-text-GGUF

lmstudio-community / gemma-3n-E4B-it-text-GGUF

DeepSeek-R1-0528-Qwen3-8B

deepseek-ai / DeepSeek-R1-0528-Qwen3-8B

8Bqwen3conversational

Deepseek-Codertext-generation1 स्पीड टेस्ट

DeepSeek-Coder-V2-Lite-Instruct-FP8

RedHatAI / DeepSeek-Coder-V2-Lite-Instruct-FP8

MoE16Bdeepseek_v2fp8vllm

Qwen3-Next-80B-A3B-Instruct

Qwen / Qwen3-Next-80B-A3B-Instruct

MoE3B active / 80B totalqwen3_nextconversational

Kimi-K3

moonshotai / Kimi-K3

2.8Tkimi_k3feature-extractioncompressed-tensors

Inkling-Small

thinkingmachines / Inkling-Small

MoE266Binkling_mm_modelconversationalaudio-text-to-text

Qwen3.5-4B-Uncensored-HauhauCS-Aggressive

HauhauCS / Qwen3.5-4B-Uncensored-HauhauCS-Aggressive

4Buncensoredqwen3.5en

सर्वश्रेष्ठ82.9 tok/s

मीडियन82.9 tok/s

न्यूनतम82.9 tok/s

Harmonic-Hermes-9B-GGUF

DJLougen / Harmonic-Hermes-9B-GGUF

9Bhermesqwen3_5reasoning

सर्वश्रेष्ठ39.2 tok/s

मीडियन39.2 tok/s

न्यूनतम39.2 tok/s

Qwen3.5-27B-Uncensored-HauhauCS-Aggressive

HauhauCS / Qwen3.5-27B-Uncensored-HauhauCS-Aggressive

Eval score· 5 evalsQ4_K_M

27Buncensoredqwen3.5en

सर्वश्रेष्ठ30.8 tok/s

मीडियन30.8 tok/s

न्यूनतम30.8 tok/s

91.5%

Mistral-Nemo-Base-2407

mistralai / Mistral-Nemo-Base-2407

12Bvllmmistral-commonen

Falcontext-generation1 स्पीड टेस्ट

Falcon-H1-Tiny-Tool-Calling-90M

tiiuae / Falcon-H1-Tiny-Tool-Calling-90M

90Mfalcon_h1edgeconversational

Falcon-H1-Tiny-Coder-90M

tiiuae / Falcon-H1-Tiny-Coder-90M

Falcontext-generation1 स्पीड टेस्ट

TinyStories-33M

roneneldan / TinyStories-33M

Gpttext-generation1 स्पीड टेस्ट

tinystories-lay4-hs128-hd2-1M

ivnle / tinystories-lay4-hs128-hd2-1M

1Mconversational

सर्वश्रेष्ठ3.6k tok/s

मीडियन3.6k tok/s

न्यूनतम3.6k tok/s

qwen-3-14b-instruct-gguf

lm-kit / qwen-3-14b-instruct-gguf

14Bconversational

सर्वश्रेष्ठ38.2 tok/s

मीडियन38.2 tok/s

न्यूनतम38.2 tok/s

12Bgemma-4gemmablereasoning

Gemmable-4-12B-MTP-GGUF

Mia-AiLab / Gemmable-4-12B-MTP-GGUF

Gemma1 स्पीड टेस्ट

सर्वश्रेष्ठ71.2 tok/s

मीडियन71.2 tok/s

न्यूनतम71.2 tok/s

Deepseek-Codertext-generation1 स्पीड टेस्ट

DeepSeek-Coder-V2-Lite-Instruct-GGUF

bartowski / DeepSeek-Coder-V2-Lite-Instruct-GGUF

imatrixconversational

सर्वश्रेष्ठ57.1 tok/s

मीडियन57.1 tok/s

न्यूनतम57.1 tok/s

Gemmaimage-text-to-text1 स्पीड टेस्ट

gemma-4-31B-v2-MLX-4bit

jdfelo / gemma-4-31B-v2-MLX-4bit

31Bmlx-vlmgemma4mlx

सर्वश्रेष्ठ33.4 tok/s

मीडियन33.4 tok/s

न्यूनतम33.4 tok/s

image-text-to-text1 स्पीड टेस्ट

Unlimited-OCR

baidu / Unlimited-OCR

MoE3Bunlimited_ocrfeature-extractionbaidu

gemma-4-26B-A4B-it-QAT-MLX-4bit

lmstudio-community / gemma-4-26B-A4B-it-QAT-MLX-4bit

Gemmaimage-text-to-text1 स्पीड टेस्ट

MoE4B active / 26B totalgemma4mlxconversational

सर्वश्रेष्ठ65.3 tok/s

मीडियन65.3 tok/s

न्यूनतम65.3 tok/s

granite-4.1-8b

ibm-granite / granite-4.1-8b

8Bgranitelanguagegranite-4.1

Optimage-text-to-text1 स्पीड टेस्ट

Step-3.7-Flash-NVFP4

stepfun-ai / Step-3.7-Flash-NVFP4

MoE104Bstep3p7vision-languagemultimodal

सर्वश्रेष्ठ27.4 tok/s

मीडियन27.4 tok/s

न्यूनतम27.4 tok/s

Qwable-3.6-35b

Mia-AiLab / Qwable-3.6-35b

35Bqwen3qwen3.6fable

सर्वश्रेष्ठ54.1 tok/s

मीडियन54.1 tok/s

न्यूनतम54.1 tok/s

20Buncensoredabliteratedmxfp4

GPT-OSS-20B-Uncensored-HauhauCS-Aggressive

HauhauCS / GPT-OSS-20B-Uncensored-HauhauCS-Aggressive

Gpt1 स्पीड टेस्ट

सर्वश्रेष्ठ66.8 tok/s

मीडियन66.8 tok/s

न्यूनतम66.8 tok/s

UncensoredLM-DeepSeek-R1-Distill-Qwen-14B

uncensoredai / UncensoredLM-DeepSeek-R1-Distill-Qwen-14B

Deepseek-R11 स्पीड टेस्ट

14Bqwen2

सर्वश्रेष्ठ27.0 tok/s

मीडियन27.0 tok/s

न्यूनतम27.0 tok/s

Qwen3.5-9B-Red_Team

LuisPPB16 / Qwen3.5-9B-Red_Team

9Bqwen3_5llama.cppunsloth

सर्वश्रेष्ठ35.9 tok/s

मीडियन35.9 tok/s

न्यूनतम35.9 tok/s

Qwentext-generation1 स्पीड टेस्ट

Qwen3.5-0.8B-Q8_0.gguf

Manojb / Qwen3.5-0.8B-Q8_0.gguf

Qwen3.6-35B-A3B-4bit-DWQ

mlx-community / Qwen3.6-35B-A3B-4bit-DWQ

MoE3B active / 35B totalmlxqwen3_5_moeconversational

सर्वश्रेष्ठ78.9 tok/s

मीडियन78.9 tok/s

न्यूनतम78.9 tok/s

Llama-2-7b

meta-llama / Llama-2-7b

7Bfacebookmetallama-2

GLM-5

zai-org / GLM-5

MoE754Bglm_moe_dsaconversationalen

Deepseektext-generation1 स्पीड टेस्ट

DeepSeek-V4-Flash-2bit-DQ

mlx-community / DeepSeek-V4-Flash-2bit-DQ

MoE284Bmlxdeepseek_v4conversational

सर्वश्रेष्ठ17.0 tok/s

मीडियन17.0 tok/s

न्यूनतम17.0 tok/s

Qwenimage-text-to-text1 स्पीड टेस्ट

Qwen3-VL-2B-Instruct

Qwen / Qwen3-VL-2B-Instruct

2Bqwen3_vlconversational

सर्वश्रेष्ठ27.9 tok/s

मीडियन27.9 tok/s

न्यूनतम27.9 tok/s

Llamaimage-text-to-text1 स्पीड टेस्ट

Gemopus-4-26B-A4B-it-GGUF

Jackrong / Gemopus-4-26B-A4B-it-GGUF

MoE4B active / 26B totalgemma4llama.cppvision

सर्वश्रेष्ठ94.5 tok/s

मीडियन94.5 tok/s

न्यूनतम94.5 tok/s

deepseek-r1-distill-qwen-14b-awq

casperhansen / deepseek-r1-distill-qwen-14b-awq

Deepseek-R1

14Bqwen24-bitawq

Ornith-1.5-9B-NVFP4

ornith-ai / Ornith-1.5-9B-NVFP4

9Bqwen3_5image-text-to-textconversational

granite-4.1-3b-base

ibm-granite / granite-4.1-3b-base

3Bgranitelanguagegranite-4.1

ZAYA1-74B-preview-legacy

Zyphra / ZAYA1-74B-preview-legacy

MoE74Bzaya

ZAYA1-base

Zyphra / ZAYA1-base

9Bzayaconversational

MiMo-V2.6-Pro-RL

XiaomiMiMo / MiMo-V2.6-Pro-RL

MoE1Tmimo_v2multimodalvision-language

Inkling

thinkingmachines / Inkling

MoE952Binkling_mm_modelconversationalaudio-text-to-text

Hy4-preview-FP8

tencent / Hy4-preview-FP8

Opttext-generation

MoE804Bhy_v4hunyuanhy4

Step-3.7-Flash-FP8

stepfun-ai / Step-3.7-Flash-FP8

MoE201Bstep3p7vision-languagemultimodal

GLM-5.3-Flash

RedHatAI / GLM-5.3-Flash

321Bglm5_nextconversationalen

GLM-5.3-Flash-BF16

RedHatAI / GLM-5.3-Flash-BF16

321Bglm5_nextimage-text-to-textconversational

Qwen3.6-35B-A3B

RedHatAI / Qwen3.6-35B-A3B

MoE3B active / 35B totalqwen3_5_moevllmconversational

Muse-Glimmer-30B

RedHatAI / Muse-Glimmer-30B

30Bmuse_glimmerconversational

Qwen3.5-27B-GPTQ-Int4

Qwen / Qwen3.5-27B-GPTQ-Int4

27Bqwen3_5conversational4-bit

GLM-4.5-Air-Base

zai-org / GLM-4.5-Air-Base

MoE110Bglm4_moeenzh

INTELLECT-2

PrimeIntellect / INTELLECT-2

33Bqwen2

Laguna-XS.2

poolside / Laguna-XS.2

MoE33Blagunalaguna-xs.2vllm

Laguna-M.1

poolside / Laguna-M.1

MoE226Bvllmlagunalaguna-m.1

limite-1b-violetto

paradigma-inc / limite-1b-violetto

1Blimitemathematicsreasoning

limite-1b-base

paradigma-inc / limite-1b-base

1Blimitemathematicsreasoning

limite-1b-base-soup

paradigma-inc / limite-1b-base-soup

1Blimitemathematicsreasoning

MoE55B active / 550B totalnemotron_hnvidianemotron-3

NVIDIA-Nemotron-3-Ultra-550B-A55B-NVFP4

nvidia / NVIDIA-Nemotron-3-Ultra-550B-A55B-NVFP4

Opttext-generation

Kimi-K2.7-Code

moonshotai / Kimi-K2.7-Code

1Tkimi_k25feature-extractioncompressed-tensors

MoE12B active / 120B totalnemotron_hnvidianemotron-3

NVIDIA-Nemotron-3-Super-120B-A12B-FP8

nvidia / NVIDIA-Nemotron-3-Super-120B-A12B-FP8

Opttext-generation

Nanbeige4.2-3B-Base

Nanbeige / Nanbeige4.2-3B-Base

3Bnanbeigellmconversational

Mistral-Small-4-119B-2603

mistralai / Mistral-Small-4-119B-2603

119Bmistral3vLLMen

Ministral-3-8B-Base-2512

mistralai / Ministral-3-8B-Base-2512

8Bvllmmistral3mistral-common

MoE8Bbailing_hybridcompressed-tensors

Ling-3.0-tiny-int4

inclusionAI / Ling-3.0-tiny-int4

Ling-3.0-tiny-fp8

inclusionAI / Ling-3.0-tiny-fp8

MoE8Bbailing_hybridfp8

Ling-3.0-tiny-base

inclusionAI / Ling-3.0-tiny-base

MoE8Bbailing_hybrid

Ling-3.0-tiny-base-midtrain

inclusionAI / Ling-3.0-tiny-base-midtrain

MoE8Bbailing_hybrid

Ling-3.0-flash-fp8

inclusionAI / Ling-3.0-flash-fp8

MoE127Bbailing_hybridconversationalfp8

Ling-3.0-flash-fp4

inclusionAI / Ling-3.0-flash-fp4

MoE66Bbailing_hybridconversational8-bit

MoE104Bbailing_hybridencompressed-tensors

Ling-2.6-flash-int4

inclusionAI / Ling-2.6-flash-int4

MoE107Bbailing_hybridenfp8

Ling-2.6-flash-fp8

inclusionAI / Ling-2.6-flash-fp8

Ling-2.6-flash-base

inclusionAI / Ling-2.6-flash-base

MoE107Bbailing_hybridconversationalfp8

dots3-note-prev-fp8

dots-studio / dots3-note-prev-fp8

MoE288Bdots3_notetext-generationdots3

lift

datalab-to / lift

10Bqwen3_5pdfextraction

219Bcohere2_visionconversationalchat

command-a-plus-05-2026-bf16

CohereLabs / command-a-plus-05-2026-bf16

Cohereimage-text-to-text

Trinity-Nano-Base-Pre-Anneal

arcee-ai / Trinity-Nano-Base-Pre-Anneal

Arceetext-generation

MoE6Bafmoeenes

MoE399Bafmoeconversationalen

Trinity-Large-TrueBase

arcee-ai / Trinity-Large-TrueBase

Arceetext-generation

Olmo-3-1025-7B

allenai / Olmo-3-1025-7B

7Bolmo3en

LFM2.5-24B-A2B-Base

LiquidAI / LFM2.5-24B-A2B-Base

MoE2B active / 24B totallfm2_moeliquidlfm2.5

MiMo-V2.6-Flash-MOPD

XiaomiMiMo / MiMo-V2.6-Flash-MOPD

MoE311Bmimo_v2multimodalvision-language

Devstral-2-123B-Instruct-2512

mistralai / Devstral-2-123B-Instruct-2512

123Bvllmministral3mistral-common

Qwen2-VL-7B

Qwen / Qwen2-VL-7B

7Bqwen2_vlmultimodalconversational

Qwen3.5-9B-NSC-ACE-SABER-GGUF

GestaltLabs / Qwen3.5-9B-NSC-ACE-SABER-GGUF

9Bqwen3_5nsc-acesaber

MoE2B active / 24B totalconversational

LFM2-24B-A2B-GGUF

lmstudio-community / LFM2-24B-A2B-GGUF