Gemini 1.5 Pro 002: Complete Benchmark Performance Guide

google 1.5-pro-002 2024-09-24 Proprietary

In-depth analysis of Gemini 1.5 Pro 002 performance on MMLU, HumanEval, GSM8K, MATH, BBH benchmarks, including pricing and use cases.

Key Specifications

Vendorgoogle
Version1.5-pro-002
Release Date2024-09-24
Context Window2e+06 tokens
Input Modalitiestext, image, audio, video
Output Modalitiestext
LicenseProprietary
Documentationhttps://ai.google.dev/gemini-api/docs

Benchmark Performance

BenchmarkScoreUnitEvaluated AtNotesSource
MMLU87%2024-09-245-shotview
HUMANEVAL90.3pass@12024-09-24—view
GSM8K95%2024-09-240-shot CoTview
MATH55.9%2024-09-240-shot CoTview
BBH87.2%2024-09-243-shot CoTview
GPQA56.5%2024-09-240-shotview
IFEVAL87.9%2024-09-24prompt_strictview
ARC95.9%2024-09-24challengeview
MUSR71.6%2024-09-240-shotview
WINOGRANDE85.8%2024-09-240-shotview

Pricing

TierPriceCurrency
Input$1.25 / MtokUSD
Output$5 / MtokUSD
Cache Read$0 / MtokUSD
Cache Write$0 / MtokUSD

Source: https://ai.google.dev/pricing · as of 2024-09-24

Compliance

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

Gemini 1.5 Pro 002

모델 개요

Google Gemini 1.5 Pro 002 生产版本, 2M 上下文, 改进指令遵循与数学推理能力。

핵심 사양

공급업체버전출시일컨텍스트 창입력 모달리티출력 모달리티라이선스
Google1.5-pro-0022024-09-242000Ktext, image, audio, videotextProprietary

벤치마크 성능

벤치마크점수단위비고
MMLU (Massive Multitask Language Understanding)87.0%5-shot
HumanEval90.3pass@1—
GSM8K (Grade School Math 8K)95.0%0-shot CoT
MATH55.9%0-shot CoT
BBH (BIG-Bench Hard)87.2%3-shot CoT
GPQA56.5%0-shot
IFEval87.9%prompt_strict
ARC95.9%challenge
MUSR71.6%0-shot
WinoGrande85.8%0-shot

가격

입력출력캐시 읽기캐시 쓰기
————

백만 토큰당

강점

  • MMLU score 87.0, strong knowledge reasoning.
  • HumanEval 90.3, excellent code generation.
  • GSM8K 95.0, robust math reasoning.
  • 支持文本、图像、音频多模态输入。
  • 上下文窗口 2000K，支持长文本。

약점

  • 闭源专有模型，不支持自托管。

사용 사례

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

참고문헌