中型开放权重 AI 模型比较(40B-150B)

参数量介于 40B 和 150B 之间的开放权重 AI 模型。

如果模型权重可供下载,我们便将其视为开放权重模型。这样用户就可以在自己的基础设施上自行托管,并通过微调等方式定制模型。

如需了解包括方法论在内的更多详情,请参阅常见问题。

NVIDIA 标志Nemotron 3.5 Super VL 和 InclusionAI 标志Ling-3.0-flash-VL 是 Artificial Analysis Progress Index v1 得分最高的中型开放权重模型,定义为参数量介于 40B 和 150B 之间的模型,其次是 InclusionAI 标志Ling-3.0-flash-Fin 和 NVIDIA 标志Nemotron 3.5 Bolt 74B-A9B。
Artificial Analysis Openness Index · Higher is better
Artificial Analysis Progress Index v1 · Higher is better
可训练参数(十亿)

开放性

Artificial Analysis Openness Index:得分

Openness Index 以 0 至 100 的标准化评分衡量模型的开放程度(分数越高,越开放)

智能

Artificial Analysis Progress Index

Artificial Analysis Progress Index v1 is an item response theory estimate of general capability across Artificial Analysis evaluations

Evaluations

Evaluations measured independently by Artificial Analysis · AA-PersonalityBench scores are percentile ranks from 1 to 100
查看更多

Agentic knowledge work, Elo

Agentic SaaS workflows

Agentic scientific research workflows in a terminal

Agentic software engineering in open-source repositories

Reasoning & knowledge

Professional document reasoning, All-pass

Agentic coding & terminal use

Physics reasoning

Quantitative analysis on spreadsheets & documents

Legal agentic work, Hallucination-Gated All-Pass Rate

No data available

Agentic business operations

Agentic real-world work tasks, Elo

Agentic interactive 3D experiences, Elo

No data available

Product and brand videos designed in code, Elo

No data available

Kubernetes incident root-cause analysis

Medical long context reasoning

Visual reasoning

Rich, varied, well-crafted language

Well-structured, scannable, answer-first

Truthful, acknowledges uncertainty, non-sycophantic

Use of filler and generic language

规模

模型规模:总参数量与活跃参数量

Comparison between total model parameters and parameters active during inference (billions)

Progress Index 与活跃参数

Artificial Analysis Progress Index v1 · Active parameters at inference time (billions)
Most attractive quadrant
Pareto line

Progress Index 与总参数量

Artificial Analysis Progress Index v1 · Size in parameters (billions)
Most attractive quadrant
Pareto line

上下文窗口

上下文窗口

Context window: tokens limit · Higher is better

更多详情

权重
服务商基准测试
Nemotron 3.5 Super VL
NVIDIA 标志NVIDIA
284
120B
推理时启用 12B 个参数
1M
$0.13
4.5
300
PrimaLabsDeepInfraBitdeer AI
+6
Ling-3.0-flash-VL
InclusionAI 标志InclusionAI
281
124B
推理时启用 5.5B 个参数
262k
-
-
-
-
Ling-3.0-flash-Fin
InclusionAI 标志InclusionAI
279
124B
推理时启用 5.1B 个参数
262k
-
-
-
-
Nemotron 3.5 Bolt 74B-A9B
NVIDIA 标志NVIDIA
269
74B
推理时启用 9B 个参数
262k
-
-
-
-
Qwen3.5 122B A10B (Reasoning)
Alibaba 标志Alibaba
261
125B
推理时启用 10B 个参数
262k
$0.68
-
102
SiliconFlowDeepInfraAlibaba CloudNovita
Mistral Medium 3.5
Mistral 标志Mistral
257
128B
256k
$1.16
4.5
145
Mistral
Qwen3.5 122B A10B (Non-reasoning)
Alibaba 标志Alibaba
250
125B
推理时启用 10B 个参数
262k
$0.68
-
93
DeepInfraAlibaba Cloud
gpt-oss-120b (High)
OpenAI 标志OpenAI
240
117B
推理时启用 5.1B 个参数
131k
$0.18
6.8
181
GroqScalewayCoreWeave
+13
Qwen3 Coder Next
Alibaba 标志Alibaba
227
79.7B
推理时启用 3B 个参数
256k
$0.36
-
61
Amazon BedrockParasail
Qwen3 Next 80B A3B (Reasoning)
Alibaba 标志Alibaba
225
80B
推理时启用 3B 个参数
262k
-
-
-
-
HyperNova 60B 2605 (High, Based on gpt-oss-120b)
Multiverse Computing 标志Multiverse Computing
224
58.7B
推理时启用 4.8B 个参数
131k
-
-
-
-
INTELLECT-3 (Based on GLM-4.5-Air)
Prime Intellect 标志Prime Intellect
221
107B
推理时启用 12B 个参数
131k
-
-
-
-