Yi Vision: Complete Benchmark Performance Guide

other yi-vision 2024-05-13 Proprietary

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

Key Specifications

Vendorother
Versionyi-vision
Release Date2024-05-13
Context Window16384 tokens
Input Modalitiestext, image
Output Modalitiestext
LicenseProprietary
Documentationhttps://huggingface.co/models

Benchmark Performance

Benchmark Score Unit Evaluated At Notes Source
MMLU 85.5 % 2024-05-13 5-shot view
HUMANEVAL 76.1 pass@1 2024-05-13 — view
GSM8K 89.9 % 2024-05-13 0-shot CoT view
MATH 58.7 % 2024-05-13 0-shot CoT view
BBH 78.3 % 2024-05-13 3-shot CoT view
GPQA 50.3 % 2024-05-13 0-shot view
IFEVAL 81.4 % 2024-05-13 prompt_strict view
ARC 95.1 % 2024-05-13 challenge view
MUSR 61.4 % 2024-05-13 0-shot view
WINOGRANDE 82.3 % 2024-05-13 0-shot view

Pricing

Tier Price Currency
Input$3 / MtokUSD
Output$3 / MtokUSD
Cache Read$0 / MtokUSD
Cache Write$0 / MtokUSD

Source: https://huggingface.co/models · as of 2024-05-13

Compliance

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

Yi Vision

Model Overview

01.AI Yi Vision 多模态模型, 16K 上下文, 支持图像理解, 中英文视觉问答能力突出。

Core Specifications

Vendor Version Release Date Context Window Input Modalities Output Modalities License
Other yi-vision 2024-05-13 16K text, image text Proprietary

Benchmark Performance

Benchmark Score Unit Notes
MMLU (Massive Multitask Language Understanding) 85.5 % 5-shot
HumanEval 76.1 pass@1 —
GSM8K (Grade School Math 8K) 89.9 % 0-shot CoT
MATH 58.7 % 0-shot CoT
BBH (BIG-Bench Hard) 78.3 % 3-shot CoT
GPQA 50.3 % 0-shot
IFEval 81.4 % prompt_strict
ARC 95.1 % challenge
MUSR 61.4 % 0-shot
WinoGrande 82.3 % 0-shot

Pricing

Input Output Cache Read Cache Write
— — — —

per million tokens

Strengths

  • MMLU score 85.5, strong knowledge reasoning.
  • GSM8K 89.9, robust math reasoning.
  • Input Modalities: text, image, audio.

Weaknesses

  • Proprietary, not self-hostable.
  • Context window 16K is limited.

Use Cases

  • Code generation and debugging
  • Vision and image understanding

References