跳轉到主要內容

Nova Pro

AmazonAmazonProprietary

描述

Amazon Nova Pro is a highly-capable multimodal model with state-of-the-art performance across text, image, and video understanding. It excels at core capabilities like language understanding, mathematical reasoning, and multimodal tasks while offering industry-leading speed and cost efficiency.

發布日期
2024-12-03
參數規模
上下文長度
300K
支援模態
image, text

能力雷達圖

29
general
16
coding
23
reasoning
30
science估算
70
agents
90
multimodal

Science 在缺少專門科學評測時使用推理能力代理估算。

排行榜排名

領域#排名分數來源
智能体与工具29
64.0
LS
代码能力榜345
20.0
AA
通用能力榜348
32.0
AA
数学推理298
22.0
AA
多模态榜12
88.0
LS
科学能力369
29.0
AA

基準測試分數 (LLM Stats)

Agents

MM-Mind2Web63.7%自報

Biology

GPQA46.9%自報

Code

HumanEval89.0%自報

Economics

FinQA77.2%自報
CRAG50.3%自報

Finance

MMLU85.9%自報

Frontend Development

VisualWebBench79.7%自報

General

ARC-C94.8%自報
IFEval92.1%自報
BFCL68.4%自報
MMMU61.7%自報

Grounding

GroundUI-1K81.4%自報

Image To Text

DocVQA93.5%自報
TextVQA81.5%自報

Language

Translation en→Set1 COMET2289.1%自報
Translation Set1→en COMET2289.0%自報
BBH86.9%自報
VATEX77.8%自報
Translation Set1→en spBleu44.4%自報
Translation en→Set1 spBleu43.4%自報
SQuALITY19.8%自報

Long Context

EgoSchema72.1%自報
LVBench41.6%自報

Math

GSM8k94.8%自報
DROP85.4%自報
MATH76.6%自報

Multimodal

ChartQA89.2%自報

AA 評測指數

Intelligence Index
13.5
Coding Index
11.0
Math Index
7.0
Math 500
0.8
Mmlu Pro
0.7
Gpqa
0.5
Ifbench
0.4
Livecodebench
0.2
Scicode
0.2
Lcr
0.2
Tau2
0.1
Aime
0.1
Aime 25
0.1
Terminalbench Hard
0.1
Hle
0.0

LLM Stats 分類評分

Structured Output
90
Code
90
Image To Text
90
Instruction Following
90
Legal
90
Math
80
Reasoning
80
Tool Calling
70
Vision
70
Finance
70
General
70
Healthcare
70
Language
70
Multimodal
70
Agents
60
Economics
60
Frontend Development
60
Biology
50
Chemistry
50
Physics
50
Search
50
Long Context
40
Summarization
20

定價

輸入價格$0.8 / 1M tokens
輸出價格$3.2 / 1M tokens
混合價格(3:1)$1.4 / 1M tokens

速度

Tokens/秒0.0 tokens/s
首Token延遲0.00s
首回答延遲0.00s

可用提供商

(LS 內部計價單位)

暫無提供商資料

外部連結