GPT-4.5 (Preview)
OpenAIGPTProprietary
描述
GPT-4.5 is OpenAI's most advanced model, offering improved reasoning, coding, and creative capabilities with faster performance and longer context handling than GPT-4. It features enhanced instruction following, reduced hallucinations, and better factual accuracy.
发布日期
2025-02-27
参数规模
—
上下文长度
—
支持模态
image, text
能力雷达图
20
general
50
coding
80
reasoning
60
science估算
60
agents
70
multimodal
Science 在缺少专门科学评测时使用推理能力代理估算。
排行榜排名
基准测试分数 (LLM Stats)
Biology
GPQA
69.5%自报
Code
HumanEval
88.0%自报
Aider-Polyglot Edit
44.9%自报
SWE-Bench Verified
38.0%自报
SWE-Lancer
37.3%自报
SWE-Lancer (IC-Diamond subset)
17.4%自报
Communication
Multi-IF
70.8%自报
TAU-bench Retail
68.4%自报
TAU-bench Airline
50.0%自报
Multi-Challenge
43.8%自报
Factuality
SimpleQA
62.5%自报
Finance
MMLU
90.8%自报
General
IFEval
88.2%自报
MMMLU
85.1%自报
MMMU
75.2%自报
Internal API instruction following (hard)
54.0%自报
Language
COLLIE
72.3%自报
Long Context
ComplexFuncBench
63.0%自报
OpenAI-MRCR: 2 needle 128k
38.5%自报
Math
GSM8k
97.0%自报
MathVista
72.3%自报
AIME 2024
36.7%自报
Multimodal
CharXiv-D
90.0%自报
CharXiv-R
55.4%自报
Reasoning
Graphwalks parents <128k
72.6%自报
Graphwalks BFS <128k
72.3%自报
AA 评测指数
Intelligence Index20.0
LLM Stats 分类评分
Finance90
Legal90
Healthcare80
Instruction Following80
Language80
Math80
Spatial Reasoning70
Structured Output70
Vision70
Writing70
Biology70
Chemistry70
General70
Multimodal70
Physics70
Tool Calling60
Communication60
Factuality60
Reasoning60
Code50
Long Context50
Frontend Development40
定价
输入价格免费
输出价格免费
混合价格(3:1)免费
速度
Tokens/秒0.0 tokens/s
首Token延迟0.00s
首回答延迟0.00s
可用提供商
(LS 内部计价单位)暂无提供商数据