Modelos de IA de pesos abertos: benchmarks e preços

Os pesos são os arquivos necessários para executar o modelo por conta própria com software e hardware adequados. Compare modelos com pesos disponíveis para download. Isso não implica licença de código aberto, uso comercial irrestrito ou hospedagem gratuita. Os preços API são do serviço hospedado; confira a licença antes de implantar.

Qual escolher?

Nenhum benchmark cobre todos os usos. Teste suas próprias perguntas e compare precisão, tempo de resposta, entradas e saídas aceitas e a conta real.

Modelos de IA

ModeloPontuação de inteligênciaCusto do benchmark (USD)Preço de entrada / 1 milhão de tokens (USD)Preço de saída / 1 milhão de tokens (USD)Velocidade de saída
Apertus 70B InstructSwiss AI Initiative2.00-0.822.92-
Apertus 8B InstructSwiss AI Initiative1.00-0.100.20-
Apriel-v1.5-15B-ThinkerServiceNow21.60-0.000.00-
Apriel-v1.6-15B-ThinkerServiceNow20.80-0.000.00-
Arctic InstructSnowflake3.00----
Cogito v2.1 (Reasoning)Deep Cogito--1.251.25-
Command ACohere7.50-2.5010.00Tok/s 58
Command-R (Mar '24)Cohere1.70-0.501.50-
Command-R+ (Apr '24)Cohere2.60-3.0015.00-
DBRX InstructDatabricks2.60----
DeepHermes 3 - Llama-3.1 8B Preview (Non-reasoning)Nous Research1.90----
DeepHermes 3 - Mistral 24B Preview (Non-reasoning)Nous Research5.00----
DeepSeek Coder V2 Lite InstructDeepSeek2.70----
DeepSeek LLM 67B Chat (V1)DeepSeek2.60----
DeepSeek R1 (Jan '25)DeepSeek18.60$4031.684.70-
DeepSeek R1 0528 (May '25)DeepSeek20.40-1.354.20-
DeepSeek R1 0528 Qwen3 8BDeepSeek10.30----
DeepSeek R1 Distill Llama 70BDeepSeek9.80-0.701.10Tok/s 27
DeepSeek R1 Distill Llama 8BDeepSeek6.20----
DeepSeek R1 Distill Qwen 1.5BDeepSeek3.30----
DeepSeek R1 Distill Qwen 14BDeepSeek9.70----
DeepSeek R1 Distill Qwen 32BDeepSeek11.00----
DeepSeek V3 (Dec '24)DeepSeek14.20$20.00.360.89-
DeepSeek V3 0324DeepSeek15.20$35.20.271.12-
DeepSeek V3.1 (Non-reasoning)DeepSeek21.40-0.561.68-
DeepSeek V3.1 (Reasoning)DeepSeek21.00-0.591.69-
DeepSeek V3.1 Terminus (Non-reasoning)DeepSeek21.70-0.271.00-
DeepSeek V3.1 Terminus (Reasoning)DeepSeek31.10-1.642.75-
DeepSeek V3.2 (Non-reasoning)DeepSeek25.10-0.280.42-
DeepSeek V3.2 (Reasoning)DeepSeek32.80-0.280.42-
DeepSeek V3.2 Exp (Non-reasoning)DeepSeek21.70-0.280.42-
DeepSeek V3.2 Exp (Reasoning)DeepSeek25.90-0.280.42-
DeepSeek V3.2 SpecialeDeepSeek22.60----
DeepSeek V4 Flash (Non-reasoning)DeepSeek29.30-0.140.28-
DeepSeek V4 Flash (Reasoning, High Effort)DeepSeek39.00$58.50.140.28-
DeepSeek V4 Flash (Reasoning, Max Effort)DeepSeek42.10$1040.130.28-
DeepSeek V4 Flash 0731 (Reasoning, Max Effort)DeepSeek51.80$3230.441.32Tok/s 114
DeepSeek V4 Pro (Non-reasoning)DeepSeek31.90-0.430.87Tok/s 88
DeepSeek V4 Pro (Reasoning, High Effort)DeepSeek43.70$1090.430.87Tok/s 87
DeepSeek V4 Pro (Reasoning, Max Effort)DeepSeek45.30$1760.430.87Tok/s 83
DeepSeek V4 Pro 0813 (Reasoning, Max Effort)DeepSeek53.20$6051.323.96Tok/s 85
DeepSeek-Coder-V2DeepSeek4.70----
DeepSeek-V2-ChatDeepSeek3.30----
DeepSeek-V2.5DeepSeek6.40----
DeepSeek-V2.5 (Dec '24)DeepSeek6.50----
Devstral Small (Jul '25)Mistral9.10----
Devstral Small (May '25)Mistral11.80----
DiffusionGemma 26B A4BGoogle13.50----
ERNIE 4.5 300B A47BBaidu8.90-0.281.10-
Exaone 4.0 1.2B (Non-reasoning)LG AI Research2.40----
Exaone 4.0 1.2B (Reasoning)LG AI Research2.50----
EXAONE 4.0 32B (Non-reasoning)LG AI Research5.70----
EXAONE 4.0 32B (Reasoning)LG AI Research10.50----
EXAONE 4.5 33BLG AI Research20.50----
EXAONE 4.5 33B (Non-reasoning)LG AI Research-----
Falcon-H1R-7BTII UAE9.70----
G9v3-39A5BAI9Stars31.60-0.000.00-
G9v3-3BAI9Stars16.20-0.000.00-
Gemma 3 1B InstructGoogle1.00-0.000.00-
Gemma 3 270MGoogle2.00----
Gemma 3 4B InstructGoogle1.00-0.000.00-
Gemma 3n E2B InstructGoogle1.00-0.000.00-
Gemma 3n E4B InstructGoogle1.00-0.060.12Tok/s 57
Gemma 3n E4B Instruct Preview (May '25)Google4.20----
Gemma 4 12B (Non-reasoning)Google13.20-0.100.30Tok/s 116
Gemma 4 12B (Reasoning)Google22.20-0.100.30Tok/s 113
Gemma 4 26B A4B (Non-reasoning)Google20.40-0.130.40Tok/s 66
Gemma 4 26B A4B (Reasoning)Google26.10$59.50.130.40-
Gemma 4 31B (Non-reasoning)Google22.30$32.80.150.40Tok/s 84
Gemma 4 E2B (Non-reasoning)Google6.20----
Gemma 4 E2B (Reasoning)Google9.50----
Gemma 4 E4B (Non-reasoning)Google8.70-0.020.10Tok/s 52
Gemma 4 E4B (Reasoning)Google12.20-0.020.10Tok/s 48
GLM-4.5 (Reasoning)Z AI19.70----
GLM-4.5-AirZ AI16.70-0.170.98Tok/s 75
GLM-4.5V (Non-reasoning)Z AI6.80-0.601.80Tok/s 100
GLM-4.5V (Reasoning)Z AI9.00-0.601.80Tok/s 85
GLM-4.6 (Non-reasoning)Z AI23.40-0.572.20Tok/s 77
GLM-4.6 (Reasoning)Z AI29.30$3140.552.20Tok/s 51
GLM-4.6V (Non-reasoning)Z AI10.90-0.300.90Tok/s 71
GLM-4.6V (Reasoning)Z AI16.90-0.300.90Tok/s 69
GLM-4.7 (Non-reasoning)Z AI27.10-0.602.20Tok/s 115
GLM-4.7 (Reasoning)Z AI34.50$6070.602.20Tok/s 113
GLM-4.7-Flash (Non-reasoning)Z AI15.60-0.070.40Tok/s 124
GLM-4.7-Flash (Reasoning)Z AI23.30-0.070.40Tok/s 98
GLM-5 (Non-reasoning)Z AI33.20-1.003.20Tok/s 40
GLM-5 (Reasoning)Z AI40.60-1.003.20Tok/s 42
GLM-5.1 (Non-reasoning)Z AI36.30-1.384.40Tok/s 74
GLM-5.1 (Reasoning)Z AI41.00$6761.384.40Tok/s 81
GLM-5.2 (max)Z AI52.60$7351.404.40Tok/s 139
GLM-5.2 (Non-reasoning)Z AI34.80-1.404.40Tok/s 140
gpt-oss-120b (high)OpenAI24.10$95.00.150.60Tok/s 176
gpt-oss-120b (low)OpenAI14.90$24.20.150.59Tok/s 202
gpt-oss-20b (high)OpenAI15.20$32.70.060.19Tok/s 125
gpt-oss-20b (low)OpenAI14.40-0.070.20Tok/s 152
Granite 3.3 8B (Non-reasoning)IBM1.30-0.030.25Tok/s 15
Granite 4.0 1BIBM1.60----
Granite 4.0 350MIBM1.00----
Granite 4.0 H 1BIBM2.20----
Granite 4.0 H 350MIBM1.00----
Granite 4.0 H SmallIBM4.90-0.060.25Tok/s 20
Granite 4.0 MicroIBM2.00----
Granite 4.1 30BIBM8.70----
Granite 4.1 3BIBM4.40----
Granite 4.1 8BIBM6.40-0.050.10Tok/s 115
Grok 2 (Dec '24)SpaceXAI7.80----
Grok-1SpaceXAI5.80----
Hermes 3 - Llama-3.1 70BNous Research4.80-0.700.70Tok/s 32
Hermes 4 - Llama-3.1 405B (Non-reasoning)Nous Research8.60-1.003.00Tok/s 29
Hermes 4 - Llama-3.1 405B (Reasoning)Nous Research8.80-1.003.00Tok/s 36
Hermes 4 - Llama-3.1 70B (Non-reasoning)Nous Research6.70-0.130.40Tok/s 94
Hermes 4 - Llama-3.1 70B (Reasoning)Nous Research9.90-0.130.40Tok/s 84
Hy3Tencent42.20$93.60.140.56Tok/s 72
Hy3-preview (Non-reasoning)Tencent26.60-0.060.21-
Hy3-preview (Reasoning)Tencent34.40-0.060.21-
HyperCLOVA X SEED Think (32B)Naver17.20----
HyperNova 60B 2605Multiverse Computing18.30----
Inkling (xhigh)Thinking Machines42.30$6981.004.05Tok/s 80
Inkling SmallThinking Machines41.20$1930.301.20Tok/s 65
INTELLECT-3Prime Intellect15.70----
Jamba 1.5 LargeAI21 Labs4.80-2.008.00-
Jamba 1.5 MiniAI21 Labs2.30-0.200.40-
Jamba 1.6 LargeAI21 Labs4.70----
Jamba 1.6 MiniAI21 Labs2.10----
Jamba 1.7 LargeAI21 Labs5.00----
Jamba 1.7 MiniAI21 Labs2.30----
Jamba Reasoning 3BAI21 Labs3.80----
K-EXAONE (Non-reasoning)LG AI Research16.90----
K-EXAONE (Reasoning)LG AI Research22.50----
K-EXAONE 2.0 0803LG AI Research31.00----
K2 Think V2MBZUAI Institute de Foundation Models17.40----
K2-V2 (high)MBZUAI Institute de Foundation Models14.20----
K2-V2 (low)MBZUAI Institute de Foundation Models8.40----
K2-V2 (medium)MBZUAI Institute de Foundation Models12.40----
Kimi K2Kimi19.70-0.572.30Tok/s 41
Kimi K2 0905Kimi24.00-0.602.50Tok/s 40
Kimi K2 ThinkingKimi33.50-0.602.50Tok/s 124
Kimi K2.5 (Non-reasoning)Kimi30.10-0.603.00Tok/s 49
Kimi K2.5 (Reasoning)Kimi36.00$3040.603.00Tok/s 48
Kimi K2.6Kimi45.10$8410.954.00Tok/s 47
Kimi K2.6 (Non-reasoning)Kimi35.40-0.954.00Tok/s 44
Kimi K2.7 CodeKimi43.00$5450.954.00Tok/s 39
Kimi K3 (low)Kimi48.30$2833.0015.00Tok/s 39
Kimi K3 (max)Kimi59.70$24253.0015.00Tok/s 39
Kimi Linear 48B A3B InstructKimi8.40----
LFM2 1.2BLiquid AI1.00----
LFM2 2.6BLiquid AI2.30----
LFM2 24B A2BLiquid AI4.60----
LFM2 8B A1BLiquid AI1.30----
LFM2.5-1.2B-InstructLiquid AI2.30----
LFM2.5-1.2B-ThinkingLiquid AI2.30----
LFM2.5-8B-A1BLiquid AI8.10-0.000.00Tok/s 338
LFM2.5-VL-1.6BLiquid AI1.00-0.000.00Tok/s 353
Ling 2.6 FlashInclusionAI14.20-0.100.30Tok/s 104
Ling 3.0 FlashInclusionAI37.80$72.80.070.22-
Ling-1TInclusionAI12.70----
Ling-2.6-1TInclusionAI26.60-0.302.50-
Ling-flash-2.0InclusionAI9.60-0.140.57Tok/s 91
Ling-mini-2.0InclusionAI3.40----
Llama 2 Chat 13BMeta2.60----
Llama 2 Chat 70BMeta2.60----
Llama 2 Chat 7BMeta3.90-0.050.25-
Llama 3 Instruct 70BMeta3.10-0.652.75-
Llama 3 Instruct 8BMeta1.00-0.040.14-
Llama 3.1 Instruct 405BMeta8.30----
Llama 3.1 Instruct 70BMeta6.50-0.560.56Tok/s 56
Llama 3.1 Instruct 8BMeta7.40-0.020.05Tok/s 133
Llama 3.1 Nemotron Instruct 70BNVIDIA7.40-1.201.20Tok/s 181
Llama 3.1 Nemotron Nano 4B v1.1 (Reasoning)NVIDIA8.40----
Llama 3.1 Nemotron Ultra 253B v1 (Reasoning)NVIDIA8.90-0.601.80Tok/s 53
Llama 3.1 Tulu3 405BAllen Institute for AI8.10----
Llama 3.2 Instruct 11B (Vision)Meta3.00-0.340.34Tok/s 44
Llama 3.2 Instruct 1BMeta1.00----
Llama 3.2 Instruct 3BMeta3.90----
Llama 3.2 Instruct 90B (Vision)Meta6.00----
Llama 3.3 Instruct 70BMeta9.30-0.660.72Tok/s 87
Llama 3.3 Nemotron Super 49B v1 (Non-reasoning)NVIDIA8.30----
Llama 3.3 Nemotron Super 49B v1 (Reasoning)NVIDIA12.20----
Llama 4 MaverickMeta14.50$35.90.270.85Tok/s 118
Llama 4 ScoutMeta10.30$11.60.180.66Tok/s 133
Llama 65BMeta1.70----
Llama Nemotron Super 49B v1.5 (Non-reasoning)NVIDIA8.50-0.400.40Tok/s 126
Llama Nemotron Super 49B v1.5 (Reasoning)NVIDIA12.40-0.400.40Tok/s 128
LongCat 2.0LongCat34.00$4420.752.95Tok/s 45
LongCat Flash LiteLongCat17.40----
Magistral Small 1Mistral10.60----
Magistral Small 1.2Mistral11.50$3010.501.50Tok/s 155
MiMo-V2-Flash (Feb 2026)Xiaomi34.00----
MiMo-V2-Flash (Non-reasoning)Xiaomi25.10----
MiMo-V2-Flash (Reasoning)Xiaomi31.90-0.100.30-
MiMo-V2.5Xiaomi38.00$25.20.140.28Tok/s 69
MiMo-V2.5-ProXiaomi42.90$98.90.430.87Tok/s 55
MiMo-V2.5-Pro (Non-reasoning)Xiaomi28.40-0.430.87Tok/s 54
MiniCPM-V 4.6 1.3BOpenBMB3.80----
MiniCPM5-1B (Non-reasoning)OpenBMB11.70----
MiniCPM5-1B (Reasoning)OpenBMB11.90----
MiniMax M1 40kMiniMax14.50----
MiniMax M1 80kMiniMax17.90-0.552.20-
MiniMax-M2MiniMax28.90-0.301.20Tok/s 101
MiniMax-M2.1MiniMax32.10-0.301.20Tok/s 99
MiniMax-M2.5MiniMax34.50-0.301.20Tok/s 105
MiniMax-M2.7MiniMax38.90$1430.301.20Tok/s 67
MiniMax-M3MiniMax45.40$2050.301.20Tok/s 109
Ministral 3 14BMistral11.20$1190.200.20Tok/s 98
Ministral 3 3BMistral7.10$91.60.100.10Tok/s 200
Ministral 3 8BMistral9.00$1140.150.15Tok/s 129
Mistral 7B InstructMistral1.70-0.250.25Tok/s 126
Mistral Large 2 (Jul '24)Mistral7.00-2.006.00-
Mistral Large 2 (Nov '24)Mistral9.00----
Mistral Large 3Mistral15.90$71.10.501.50Tok/s 52
Mistral Medium 3.5Mistral30.40$8831.507.50Tok/s 151
Mistral Small (Sep '24)Mistral4.30-0.200.60Tok/s 158
Mistral Small 3Mistral6.70-0.100.30Tok/s 163
Mistral Small 3.1Mistral14.90$29.80.100.30Tok/s 153
Mistral Small 3.2Mistral10.70$1290.100.30Tok/s 157
Mistral Small 4 (Non-reasoning)Mistral12.30-0.150.60Tok/s 157
Mistral Small 4 (Reasoning)Mistral19.70$87.70.150.60Tok/s 175
Mixtral 8x22B InstructMistral4.00----
Mixtral 8x7B InstructMistral2.00-0.450.70-
Molmo 7B-DAllen Institute for AI3.40----
Molmo2-8BAllen Institute for AI1.60----
Motif 3Motif Technologies47.40----
Muse Glimmer (high)Meta35.10$1050.321.35Tok/s 104
Nanbeige4.1-3BNanbeige11.00----
Nemotron 3 Nano Omni 30B A3B ReasoningNVIDIA15.00-0.070.30Tok/s 328
Nemotron 3 Super 120B A12B (Reasoning)NVIDIA25.70$2540.200.80Tok/s 140
Nemotron 3 Ultra 550B A55B (Reasoning)NVIDIA38.30$5340.602.75Tok/s 154
Nemotron 3.5 LightningNVIDIA23.60$59.70.050.20Tok/s 300
Nemotron Cascade 2 30B A3BNVIDIA17.80----
Nex-N2-ProNex AGI41.70-0.502.50Tok/s 138
NVIDIA Nemotron 3 Nano 30B A3B (Non-reasoning)NVIDIA7.20-0.050.20Tok/s 190
NVIDIA Nemotron 3 Nano 30B A3B (Reasoning)NVIDIA14.50$39.40.050.20Tok/s 227
NVIDIA Nemotron 3 Nano 4BNVIDIA8.60----
NVIDIA Nemotron Nano 12B v2 VL (Non-reasoning)NVIDIA4.20-0.200.60Tok/s 200
NVIDIA Nemotron Nano 12B v2 VL (Reasoning)NVIDIA8.80-0.200.60Tok/s 137
NVIDIA Nemotron Nano 9B V2 (Non-reasoning)NVIDIA7.20-0.050.20Tok/s 168
NVIDIA Nemotron Nano 9B V2 (Reasoning)NVIDIA8.70-0.040.16Tok/s 173
OLMo 2 32BAllen Institute for AI4.70----
OLMo 2 7BAllen Institute for AI3.50----
Olmo 3 32B ThinkAllen Institute for AI6.10----
Olmo 3 7B InstructAllen Institute for AI2.40-0.100.20-
Olmo 3 7B ThinkAllen Institute for AI3.60----
Olmo 3.1 32B InstructAllen Institute for AI6.20----
Olmo 3.1 32B ThinkAllen Institute for AI7.90-0.000.00-
OpenChat 3.5 (1210)OpenChat2.60----
Phi-3 Mini Instruct 3.8BMicrosoft4.30----
Phi-4Microsoft4.60-0.130.50Tok/s 23
Phi-4 Mini InstructMicrosoft5.70-0.000.00Tok/s 44
Phi-4 Multimodal InstructMicrosoft4.20-0.000.00Tok/s 18
Pixtral LargeMistral8.00----
Qwen Chat 14BAlibaba1.70----
Qwen Chat 72BAlibaba3.00----
Qwen1.5 Chat 110BAlibaba3.70----
Qwen2 Instruct 72BAlibaba5.70----
Qwen2.5 Coder Instruct 32BAlibaba6.90----
Qwen2.5 Coder Instruct 7B Alibaba4.10----
Qwen2.5 Instruct 32BAlibaba7.20----
Qwen2.5 Instruct 72BAlibaba9.40-0.470.49-
Qwen3 0.6B (Non-reasoning)Alibaba1.00----
Qwen3 0.6B (Reasoning)Alibaba1.00----
Qwen3 1.7B (Non-reasoning)Alibaba1.10----
Qwen3 1.7B (Reasoning)Alibaba2.20----
Qwen3 14B (Non-reasoning)Alibaba6.80-0.351.40Tok/s 64
Qwen3 14B (Reasoning)Alibaba10.40-0.354.20Tok/s 61
Qwen3 235B A22B (Non-reasoning)Alibaba10.80-0.702.80Tok/s 62
Qwen3 235B A22B (Reasoning)Alibaba13.50-0.708.40Tok/s 59
Qwen3 235B A22B 2507 (Reasoning)Alibaba19.90$6100.708.40Tok/s 65
Qwen3 235B A22B 2507 InstructAlibaba18.40-0.702.80Tok/s 61
Qwen3 30B A3B (Non-reasoning)Alibaba6.60-0.200.80Tok/s 107
Qwen3 30B A3B (Reasoning)Alibaba9.20-0.202.40Tok/s 106
Qwen3 30B A3B 2507 (Reasoning)Alibaba14.60$1520.202.40Tok/s 148
Qwen3 30B A3B 2507 InstructAlibaba8.90-0.200.80Tok/s 141
Qwen3 32B (Non-reasoning)Alibaba8.50-0.702.80Tok/s 107
Qwen3 32B (Reasoning)Alibaba11.40-0.708.40Tok/s 104
Qwen3 4B (Non-reasoning)Alibaba6.50----
Qwen3 4B (Reasoning)Alibaba8.20----
Qwen3 4B 2507 (Reasoning)Alibaba11.90----
Qwen3 4B 2507 InstructAlibaba6.90----
Qwen3 8B (Non-reasoning)Alibaba4.80-0.180.70Tok/s 40
Qwen3 8B (Reasoning)Alibaba8.30-0.182.10Tok/s 38
Qwen3 Coder 30B A3B InstructAlibaba13.60-0.452.25Tok/s 95
Qwen3 Coder 480B A35B InstructAlibaba18.20-1.507.50Tok/s 73
Qwen3 Coder NextAlibaba21.30$2500.351.20Tok/s 151
Qwen3 Next 80B A3B (Reasoning)Alibaba16.90$4000.506.00Tok/s 199
Qwen3 Next 80B A3B InstructAlibaba13.80-0.502.00Tok/s 191
Qwen3 Omni 30B A3B (Reasoning)Alibaba9.50-0.250.97Tok/s 99
Qwen3 Omni 30B A3B InstructAlibaba4.80-0.250.97Tok/s 93
Qwen3 VL 235B A22B (Reasoning)Alibaba20.90-0.708.40Tok/s 55
Qwen3 VL 235B A22B InstructAlibaba14.40-0.702.80Tok/s 50
Qwen3 VL 30B A3B (Reasoning)Alibaba13.40-0.202.40Tok/s 112
Qwen3 VL 30B A3B InstructAlibaba9.90-0.200.80Tok/s 112
Qwen3 VL 32B (Reasoning)Alibaba18.10-0.708.40Tok/s 90
Qwen3 VL 32B InstructAlibaba11.00-0.702.80Tok/s 66
Qwen3 VL 4B (Reasoning)Alibaba7.70----
Qwen3 VL 4B InstructAlibaba3.70----
Qwen3 VL 8B (Reasoning)Alibaba10.50-0.182.10Tok/s 112
Qwen3 VL 8B InstructAlibaba8.20-0.180.70Tok/s 119
Qwen3.5 0.8B (Non-reasoning)Alibaba2.90----
Qwen3.5 0.8B (Reasoning)Alibaba5.20----
Qwen3.5 122B A10B (Non-reasoning)Alibaba28.20$2210.403.20Tok/s 148
Qwen3.5 122B A10B (Reasoning)Alibaba32.80$4440.403.20Tok/s 136
Qwen3.5 27B (Non-reasoning)Alibaba30.00-0.302.40Tok/s 83
Qwen3.5 27B (Reasoning)Alibaba34.60-0.302.40Tok/s 78
Qwen3.5 2B (Non-reasoning)Alibaba5.30----
Qwen3.5 2B (Reasoning)Alibaba7.40----
Qwen3.5 35B A3B (Non-reasoning)Alibaba24.30$2280.252.00Tok/s 170
Qwen3.5 35B A3B (Reasoning)Alibaba29.90-0.252.00Tok/s 151
Qwen3.5 397B A17B (Non-reasoning)Alibaba32.70-0.603.60Tok/s 95
Qwen3.5 397B A17B (Reasoning)Alibaba34.30$5270.603.60Tok/s 97
Qwen3.5 4B (Non-reasoning)Alibaba16.10-0.030.15Tok/s 52
Qwen3.5 4B (Reasoning)Alibaba20.40-0.030.15Tok/s 43
Qwen3.5 9B (Non-reasoning)Alibaba20.60-0.170.25Tok/s 89
Qwen3.5 9B (Reasoning)Alibaba21.80$2400.140.20Tok/s 81
Qwen3.6 27B (Non-reasoning)Alibaba31.30$3590.603.60Tok/s 58
Qwen3.6 27B (Reasoning)Alibaba37.70$6670.603.60Tok/s 61
Qwen3.6 35B A3B (Non-reasoning)Alibaba24.60$5300.382.25Tok/s 146
Qwen3.6 35B A3B (Reasoning)Alibaba32.10$5050.382.25Tok/s 131
Qwen3.8 2.4T A95BAlibaba57.70$16722.006.00Tok/s 48
QwQ 32BAlibaba13.40-0.661.00-
QwQ 32B-PreviewAlibaba9.10----
R1 1776Perplexity6.00----
Reka Flash 3Reka AI3.70-0.200.80-
Ring-1TInclusionAI16.30----
Ring-2.6-1TInclusionAI31.70$4550.302.50-
Ring-flash-2.0InclusionAI8.00-0.140.57-
Sarvam 105B (high)Sarvam11.90-0.040.17-
Sarvam 30B (high)Sarvam6.40-0.030.11-
Sarvam M (Reasoning)Sarvam2.60-0.000.00-
Seed-OSS-36B-InstructByteDance Seed18.50-0.210.57Tok/s 24
Solar MiniUpstage6.00-0.150.15-
Solar Open 100B (Reasoning)Upstage15.20----
Solar Open2 250BUpstage37.40----
Step 3.5 FlashStepFun26.00-0.100.30Tok/s 225
Step 3.7 FlashStepFun30.90$3200.201.15Tok/s 150
Step3 VL 10BStepFun9.30----
Tiny Aya GlobalCohere1.00-0.000.00-
Tri-21B-ThinkTrillion Labs12.30----
Tri-21B-think PreviewTrillion Labs13.60----
Trinity Large ThinkingArcee AI18.70$2550.230.88Tok/s 202

Guias

Como comparamos os modelos

A inteligência é medida pelo Artificial Analysis Intelligence Index. O custo do benchmark é o custo total em USD para executar o teste, incluindo tokens de raciocínio; não é o preço da API por token. Todas as versões usam os mesmos dados e cálculos. A pontuação geral não mede o desempenho em um idioma ou tarefa específicos.