Comparação de modelos de pesos abertos

Comparação e análise de modelos de IA de pesos abertos em métricas importantes de desempenho, como qualidade, desempenho, velocidade de inferência, janela de contexto, número de parâmetros e detalhes da licença.

Os modelos são considerados de pesos abertos (também chamados frequentemente de código aberto) quando seus pesos estão disponíveis para download. Isso permite a auto-hospedagem em sua própria infraestrutura e a personalização do modelo, por exemplo, por meio de ajuste fino.

Para mais detalhes sobre nossa metodologia, consulte as perguntas frequentes.

Logo: KimiKimi K3 (max) e Logo: Z AIGLM-5.2 (max) são os modelos com maior inteligência entre os modelos de pesos abertos, seguidos por Logo: DeepSeekDeepSeek V4 Flash 0731 (max) e Logo: KimiKimi K3 (low).

Destaques

Abertura

Artificial Analysis Openness Index · Higher is better

Inteligência

Artificial Analysis Intelligence Index · Higher is better

Parâmetros totais

Parâmetros treináveis em bilhões

Abertura

Artificial Analysis Openness Index: Score

Openness Index assesses model openness on a 0 to 100 normalized scale (higher is more open)
Reasoning models are indicated by a lightbulb icon

Progresso dos modelos de pesos abertos

Progress in Open Weights vs. Proprietary Intelligence

Artificial Analysis Intelligence Index v4.1 incorporates 9 evaluations: GDPval-AA v2, 𝜏³-Banking, Terminal-Bench v2.1, SciCode, Humanity's Last Exam, GPQA Diamond, CritPt, AA-Omniscience, AA-LCR
Reasoning models are indicated by a lightbulb icon

Artificial Analysis Intelligence Index v4.1 includes: GDPval-AA v2, 𝜏³-Banking, Terminal-Bench v2.1, SciCode, Humanity's Last Exam, GPQA Diamond, CritPt, AA-Omniscience, AA-LCR. See Intelligence Index methodology for further details, including a breakdown of each evaluation and how we run them.

Indicates whether the model weights are available. Models are labelled as 'Commercial Use Restricted' if the weights are available but commercial use is limited (typically requires obtaining a paid license).

Evolução da inteligência dos modelos de linguagem de pesos abertos por laboratório

Reasoning models are indicated by a lightbulb icon

Artificial Analysis Intelligence Index v4.1 includes: GDPval-AA v2, 𝜏³-Banking, Terminal-Bench v2.1, SciCode, Humanity's Last Exam, GPQA Diamond, CritPt, AA-Omniscience, AA-LCR. See Intelligence Index methodology for further details, including a breakdown of each evaluation and how we run them.

Evolução da inteligência dos modelos de pesos abertos por tamanho

Artificial Analysis Intelligence Index v4.1 incorporates 9 evaluations: GDPval-AA v2, 𝜏³-Banking, Terminal-Bench v2.1, SciCode, Humanity's Last Exam, GPQA Diamond, CritPt, AA-Omniscience, AA-LCR
Reasoning models are indicated by a lightbulb icon

Artificial Analysis Intelligence Index v4.1 includes: GDPval-AA v2, 𝜏³-Banking, Terminal-Bench v2.1, SciCode, Humanity's Last Exam, GPQA Diamond, CritPt, AA-Omniscience, AA-LCR. See Intelligence Index methodology for further details, including a breakdown of each evaluation and how we run them.

  • Tiny: Less than or equal to 4B parameters. These are usually the smallest models in terms of resource demand.
  • Small: Less than 40B parameters.
  • Medium: Between 40B-150B parameters.
  • Large: Over 150B parameters.

Inteligência

Artificial Analysis Intelligence Index

Artificial Analysis Intelligence Index v4.1 incorporates 9 evaluations: GDPval-AA v2, 𝜏³-Banking, Terminal-Bench v2.1, SciCode, Humanity's Last Exam, GPQA Diamond, CritPt, AA-Omniscience, AA-LCR
Reasoning models are indicated by a lightbulb icon

Artificial Analysis Intelligence Index v4.1 includes: GDPval-AA v2, 𝜏³-Banking, Terminal-Bench v2.1, SciCode, Humanity's Last Exam, GPQA Diamond, CritPt, AA-Omniscience, AA-LCR. See Intelligence Index methodology for further details, including a breakdown of each evaluation and how we run them.

Intelligence Evaluations

Intelligence evaluations measured independently by Artificial Analysis · Higher is better
GDPval-AA v2

Agentic real-world work tasks, (Elo-500)/2000

𝜏³-Banking

Agentic tool use

Terminal-Bench v2.1

Agentic coding & terminal use

SciCode

Coding

Humanity's Last Exam

Reasoning & knowledge

GPQA Diamond

Scientific reasoning

CritPt

Physics reasoning

AA-Omniscience Accuracy

Knowledge

AA-Omniscience Non-Hallucination Rate

1 - hallucination rate

AA-LCR

Long context reasoning

AA-Briefcase

Agentic knowledge work, Elo

AutomationBench-AA

Agentic SaaS workflows

Harvey LAB-AA

Legal agentic work, criterion pass rate

EnterpriseOps-Gym-AANew

Agentic business operations

IFBench

Instruction following

APEX-Agents-AA

Long-horizon agentic tasks

ITBench-AA

Kubernetes incident root-cause analysis

MMMU-Pro

Visual reasoning

Reasoning models are indicated by a lightbulb icon

While model intelligence generally translates across use cases, specific evaluations may be more relevant for certain use cases.

Artificial Analysis Intelligence Index v4.1 includes: GDPval-AA v2, 𝜏³-Banking, Terminal-Bench v2.1, SciCode, Humanity's Last Exam, GPQA Diamond, CritPt, AA-Omniscience, AA-LCR. See Intelligence Index methodology for further details, including a breakdown of each evaluation and how we run them.

Tamanho

Intelligence Index por tamanho do modelo

Artificial Analysis Intelligence Index v4.1 incorporates 9 evaluations: GDPval-AA v2, 𝜏³-Banking, Terminal-Bench v2.1, SciCode, Humanity's Last Exam, GPQA Diamond, CritPt, AA-Omniscience, AA-LCR
Reasoning models are indicated by a lightbulb icon

Artificial Analysis Intelligence Index v4.1 includes: GDPval-AA v2, 𝜏³-Banking, Terminal-Bench v2.1, SciCode, Humanity's Last Exam, GPQA Diamond, CritPt, AA-Omniscience, AA-LCR. See Intelligence Index methodology for further details, including a breakdown of each evaluation and how we run them.

Model Size: Total and Active Parameters

Comparison between total model parameters and parameters active during inference
Reasoning models are indicated by a lightbulb icon

The total number of trainable weights and biases in the model, expressed in billions. These parameters are learned during training and determine the model's ability to process and generate responses.

The number of parameters actually executed during each inference forward pass, expressed in billions. For Mixture of Experts (MoE) models, a routing mechanism selects a subset of experts per token, resulting in fewer active than total parameters. Dense models use all parameters, so active equals total.

Intelligence Index vs. Active Parameters

Artificial Analysis Intelligence Index · Active parameters at inference time
Most attractive quadrant
Pareto line
Reasoning models are indicated by a lightbulb icon

Artificial Analysis Intelligence Index v4.1 includes: GDPval-AA v2, 𝜏³-Banking, Terminal-Bench v2.1, SciCode, Humanity's Last Exam, GPQA Diamond, CritPt, AA-Omniscience, AA-LCR. See Intelligence Index methodology for further details, including a breakdown of each evaluation and how we run them.

The number of parameters actually executed during each inference forward pass, expressed in billions. For Mixture of Experts (MoE) models, a routing mechanism selects a subset of experts per token, resulting in fewer active than total parameters. Dense models use all parameters, so active equals total.

Intelligence Index vs. Total Parameters

Artificial Analysis Intelligence Index · Size in parameters (billions)
Most attractive quadrant
Pareto line
Reasoning models are indicated by a lightbulb icon

Artificial Analysis Intelligence Index v4.1 includes: GDPval-AA v2, 𝜏³-Banking, Terminal-Bench v2.1, SciCode, Humanity's Last Exam, GPQA Diamond, CritPt, AA-Omniscience, AA-LCR. See Intelligence Index methodology for further details, including a breakdown of each evaluation and how we run them.

The total number of trainable weights and biases in the model, expressed in billions. These parameters are learned during training and determine the model's ability to process and generate responses.

Janela de contexto

Context Window

Context window: tokens limit · Higher is better
Reasoning models are indicated by a lightbulb icon

Larger context windows are relevant to RAG (Retrieval Augmented Generation) LLM workflows which typically involve reasoning and information retrieval of large amounts of data.

Maximum number of combined input & output tokens. Output tokens commonly have a significantly lower limit (varied by model).

Mais detalhes

Pesos
Benchmarks de provedores
Kimi K3 (max)
Logo: KimiKimi
57
2.8T
104B ativos durante a inferência
1M
$2.3
37
🤗
ModalNebiusTogether AI
+8
Ver
GLM-5.2 (max)
Logo: Z AIZ AI
51
753B
40B ativos durante a inferência
1M
$0.9
150
🤗
Blackbox AIGMIWafer
+13
Ver
DeepSeek V4 Flash 0731 (Reasoning, Max Effort)
Logo: DeepSeekDeepSeek
50
284B
13B ativos durante a inferência
1M
$0.1
104
🤗
ParasailDeepInfraDeepSeek
+3
Ver
MiniMax-M3
Logo: MiniMaxMiniMax
44
428B
23B ativos durante a inferência
1M
$0.2
73
🤗
NovitaGMICoreWeave
+7
Ver
MiMo-V2.5-Pro
Logo: XiaomiXiaomi
42
1.0T
42B ativos durante a inferência
1M
$0.2
68
🤗
DeepInfraGMINovita
+2
Ver
Inkling (xhigh)
Logo: Thinking MachinesThinking Machines
41
975B
41B ativos durante a inferência
1M
$0.7
80
🤗
Self-hostedThinking MachinesDeepInfraTogether AI
Ver
Nemotron 3 Ultra 550B A55B (Reasoning)
Logo: NVIDIANVIDIA
38
550B
55B ativos durante a inferência
262k
$0.5
145
Não disponível
GMICoreWeaveLightning AI
+6
Ver
Mistral Medium 3.5
Logo: MistralMistral
30
128B
256k
$1.2
140
🤗
Self-hostedMistral
Ver
Gemma 4 31B (Reasoning)
Logo: GoogleGoogle
29
30.7B
256k
-
35
🤗
Together AISiliconFlowFriendliAI
+10
Ver
gpt-oss-120b (high)
Logo: OpenAIOpenAI
24
117B
5.1B ativos durante a inferência
131k
$0.2
185
🤗
Self-hostedMicrosoft AzureFireworks
+18
Ver
Command A+
Logo: CohereCohere
23
218B
25B ativos durante a inferência
192k
-
205
🤗
Cohere
Ver