intersect-by-codenteam.smaller

GPT-4o mini Search Preview

87.4
Score

Overall Performance Score

OpenAI Logo OpenAI
2024-10-31
88%
TextGeneration
86%
Reasoning
85%
Coding

Overview

What is GPT-4o mini Search Preview?

Compact GPT-4o variant with web search, optimized for speed and cost-efficiency with current information access.

Created by:

OpenAI

Release Date:

2024-10-31

Capabilities Overview

TextGeneration 88%
Reasoning 86%
Coding 85%
Multimodal 87%
Safety 91%

Technical Specifications

Architecture

type: Efficient Multimodal Transformer with Search
parameters: Unknown
context: 128,000 tokens
trainingDataUpTo: October 2023 + Real-time Search
architecture: Optimized GPT-4o mini with integrated web search, efficient search processing, and balanced performance

Performance Metrics

MMLU: 88.5%
HumanEval: 82.7%
Search Speed: Fast
Response Time: 250ms+

Performance Dashboard

TextGeneration

88%

Reasoning

86%

Coding

85%

Multimodal

87%

Safety

91%

Technical Metrics

Parameters: Unknown
ContextWindow: 128000
Latency: 250
Accuracy: 88.5
Cost: $0.020/1K tokens

Benchmark Performance

MMLU 88.5%
HumanEval 82.7%
Search Speed Fast
Response Time 250ms+

Features

Web Search

Real-time web search capabilities

Fast

Faster than GPT-4o Search

Affordable

More cost-effective with search

Current Info

Access to up-to-date information

Multimodal

Text and image support

Large Context

128K token window

Pros & Cons

Advantages

  • Real-time search
  • Fast
  • Cost-effective
  • Good for simple research

Disadvantages

  • Less capable than full version
  • Preview status
  • Search quality varies

What can it do?

Quick Lookups

Fast searches for current information and facts

News Summaries

Summarize current news and events efficiently

Trend Analysis

Track and analyze current trends and topics

Frequently Asked Questions