Gemini 1.5 Flash 002: Complete Benchmark Performance Guide

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

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

Key Specifications

Vendorgoogle
Version1.5-flash-002
Release Date2024-09-24
Context Window1e+06 tokens
Input Modalitiestext, image, audio, video
Output Modalitiestext
LicenseProprietary
Documentationhttps://ai.google.dev/gemini-api/docs

Benchmark Performance

Benchmark Score Unit Evaluated At Notes Source
MMLU 81.1 % 2024-09-24 5-shot view
HUMANEVAL 66.2 pass@1 2024-09-24 — view
GSM8K 78.8 % 2024-09-24 0-shot CoT view
MATH 40.5 % 2024-09-24 0-shot CoT view
BBH 79.8 % 2024-09-24 3-shot CoT view
GPQA 32 % 2024-09-24 0-shot view
IFEVAL 77.2 % 2024-09-24 prompt_strict view
ARC 91.8 % 2024-09-24 challenge view
MUSR 59.8 % 2024-09-24 0-shot view
WINOGRANDE 81.9 % 2024-09-24 0-shot view

Pricing

Tier Price Currency
Input$0.075 / MtokUSD
Output$0.3 / 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 Flash 002

Model Overview

Google Gemini 1.5 Flash 002 生产版本, 1M 上下文, 改进质量与稳定性。

Core Specifications

Vendor Version Release Date Context Window Input Modalities Output Modalities License
Google 1.5-flash-002 2024-09-24 1000K text, image, audio, video text Proprietary

Benchmark Performance

Benchmark Score Unit Notes
MMLU (Massive Multitask Language Understanding) 81.1 % 5-shot
HumanEval 66.2 pass@1 —
GSM8K (Grade School Math 8K) 78.8 % 0-shot CoT
MATH 40.5 % 0-shot CoT
BBH (BIG-Bench Hard) 79.8 % 3-shot CoT
GPQA 32.0 % 0-shot
IFEval 77.2 % prompt_strict
ARC 91.8 % challenge
MUSR 59.8 % 0-shot
WinoGrande 81.9 % 0-shot

Pricing

Input Output Cache Read Cache Write
— — — —

per million tokens

Strengths

  • MMLU score 81.1, strong knowledge reasoning.
  • Input Modalities: text, image, audio.
  • Context window 1000K.

Weaknesses

  • Proprietary, not self-hostable.

Use Cases

  • Long document summarization
  • Vision and image understanding

References