Compare AI models

Specifications
Parameters
—6B
Context window
32k32k
Benchmarks
HumanEval
74.4%9.8%
MATH
53.2%6.9%
HellaSwag
87.8%57%
MMLU
83.7%45.2%
GSM8K
88.9%25.9%
Overview
CompanyGoogleZ.ai
Release dateFeb 8 2024Jun 25 2023
AccessClosedOpen Weight
Model detailsView modelView model

Frequently asked questions

Gemini 1.0 Ultra leads ChatGLM2-6B on 5 of the 5 benchmarks they both report. ChatGLM2-6B shipped 228 days before Gemini 1.0 Ultra, so benchmark comparisons should account for the intervening progress.

Context windows are 32k (Gemini 1.0 Ultra) vs 32k (ChatGLM2-6B). Gemini 1.0 Ultra is closed, while ChatGLM2-6B is open weight.

On HumanEval, Gemini 1.0 Ultra leads at 74.4% vs ChatGLM2-6B at 9.8%. On MATH, Gemini 1.0 Ultra leads at 53.2% vs ChatGLM2-6B at 6.9%. On HellaSwag, Gemini 1.0 Ultra leads at 87.8% vs ChatGLM2-6B at 57%. On MMLU, Gemini 1.0 Ultra leads at 83.7% vs ChatGLM2-6B at 45.2%. On GSM8K, Gemini 1.0 Ultra leads at 88.9% vs ChatGLM2-6B at 25.9%.