GLM-4V 9B: Complete Benchmark Performance Guide

other glm-4v-9b 2024-06-05 GLM-4 License

In-depth analysis of GLM-4V 9B performance on MMLU, HumanEval, GSM8K, MATH, BBH benchmarks, including pricing and use cases.

Key Specifications

Vendorother
Versionglm-4v-9b
Release Date2024-06-05
Context Window131072 tokens
Input Modalitiestext, image
Output Modalitiestext
LicenseGLM-4 License
Documentationhttps://huggingface.co/models

Benchmark Performance

Benchmark Score Unit Evaluated At Notes Source
MMLU 73.4 % 2024-06-05 5-shot view
HUMANEVAL 66 pass@1 2024-06-05 — view
GSM8K 73.5 % 2024-06-05 0-shot CoT view
MATH 33.3 % 2024-06-05 0-shot CoT view
BBH 65.1 % 2024-06-05 3-shot CoT view
GPQA 28.3 % 2024-06-05 0-shot view
IFEVAL 67.4 % 2024-06-05 prompt_strict view
ARC 90.2 % 2024-06-05 challenge view
MUSR 38.5 % 2024-06-05 0-shot view
WINOGRANDE 73.1 % 2024-06-05 0-shot view

Pricing

Tier Price Currency
Input$0.18 / MtokUSD
Output$0.18 / MtokUSD
Cache Read$0 / MtokUSD
Cache Write$0 / MtokUSD

Source: https://huggingface.co/models · as of 2024-06-05

Compliance

  • Data Residency: self-host
  • SOC2: ✗
  • HIPAA: ✗
  • GDPR: ✗
  • ISO 27001: ✗

GLM-4V 9B

Model Overview

清华智谱 GLM-4V 9B 多模态开源模型, 131K 上下文, 9B 参数, 视觉理解能力突出。

Core Specifications

Vendor Version Release Date Context Window Input Modalities Output Modalities License
Other glm-4v-9b 2024-06-05 131K text, image text GLM-4 License

Benchmark Performance

Benchmark Score Unit Notes
MMLU (Massive Multitask Language Understanding) 73.4 % 5-shot
HumanEval 66.0 pass@1 —
GSM8K (Grade School Math 8K) 73.5 % 0-shot CoT
MATH 33.3 % 0-shot CoT
BBH (BIG-Bench Hard) 65.1 % 3-shot CoT
GPQA 28.3 % 0-shot
IFEval 67.4 % prompt_strict
ARC 90.2 % challenge
MUSR 38.5 % 0-shot
WinoGrande 73.1 % 0-shot

Pricing

Input Output Cache Read Cache Write
— — — —

per million tokens

Strengths

  • Input Modalities: text, image, audio.

Weaknesses

  • Proprietary, not self-hostable.

Use Cases

  • Vision and image understanding

References