GPT-4 Vision: Complete Benchmark Performance Guide

openai 4-vision 2024-01-25 Proprietary

In-depth analysis of GPT-4 Vision performance on MMLU, HumanEval, GSM8K, MATH, BBH benchmarks, including pricing and use cases.

Key Specifications

Vendoropenai
Version4-vision
Release Date2024-01-25
Context Window128000 tokens
Input Modalitiestext, image
Output Modalitiestext
LicenseProprietary
Documentationhttps://platform.openai.com/docs/models

Benchmark Performance

BenchmarkScoreUnitEvaluated AtNotesSource
MMLU83.9%2024-01-255-shotview
HUMANEVAL75.8pass@12024-01-25—view
GSM8K83.6%2024-01-250-shot CoTview
MATH48.3%2024-01-250-shot CoTview
BBH80.9%2024-01-253-shot CoTview
GPQA41.8%2024-01-250-shotview
IFEVAL73.5%2024-01-25prompt_strictview
ARC93.3%2024-01-25challengeview
MUSR51.2%2024-01-250-shotview
WINOGRANDE86.9%2024-01-250-shotview

Pricing

TierPriceCurrency
Input$10 / MtokUSD
Output$30 / MtokUSD
Cache Read$0 / MtokUSD
Cache Write$0 / MtokUSD

Source: https://openai.com/api/pricing/ · as of 2024-01-25

Compliance

  • Data Residency: US
  • SOC2: ✓
  • HIPAA: ✓
  • GDPR: ✓
  • ISO 27001: ✓

GPT-4 Vision

Descripción del modelo

OpenAI GPT-4 Vision 正式版, 支持图像理解与多模态推理, 128K 上下文窗口。

Especificaciones principales

ProveedorVersiónFecha de lanzamientoVentana de contextoModalidades de entradaModalidades de salidaLicencia
Openai4-vision2024-01-25128Ktext, imagetextProprietary

Rendimiento en benchmarks

BenchmarkPuntuaciónUnidadNotas
MMLU (Massive Multitask Language Understanding)83.9%5-shot
HumanEval75.8pass@1—
GSM8K (Grade School Math 8K)83.6%0-shot CoT
MATH48.3%0-shot CoT
BBH (BIG-Bench Hard)80.9%3-shot CoT
GPQA41.8%0-shot
IFEval73.5%prompt_strict
ARC93.3%challenge
MUSR51.2%0-shot
WinoGrande86.9%0-shot

Precios

EntradaSalidaLectura cachéEscritura caché
————

por millón de tokens

Fortalezas

  • MMLU score 83.9, strong knowledge reasoning.
  • 支持文本、图像、音频多模态输入。

Debilidades

  • 闭源专有模型，不支持自托管。

Casos de uso

  • 代码生成与调试
  • 视觉与图像理解
  • Agent 工作流与工具调用

Referencias