Comparison in capability with GSM8K benchmark for major generative AI programs 2023

Published by
Bergur Thormundsson,
Feb 16, 2024
OpenAI's GPT-4 and Anthropic's Claude 2 are the most capable large language models when compared with the GSM8k benchmark in 2023. The rapid increase in AI capabilities is noticeable when comparing the scores of GPT-3.5 and GPT-4, where the latter iteration scored 35 percent higher than the former.

GSM8k benchmark comparison between major generative artificial intelligence (AI) programs in 2023

© Statista 2024
Download
Show detailed source information?
Register for free
Already a member?
Log in
Source

Use Ask Statista Research Service

Release date

November 2023

Region

Worldwide

Survey time period

2023

Supplementary notes

Source clarifies: "GSM8k: Middle school math word problems, (Cobbe et al. 2021), using the chain-of-thought prompt."

Citation formats
Access all statistics starting from $2,388 USD yearly *

* For commercial use only

Basic Account

For single users
$0 USD
Always free
Access limited to Free Statistics. Premium Statistics are not included.
  • Free Statistics
Based on your interests

Starter Account

For single users
$199 USD
per month, billed annually 1
Buy now
  • Free Statistics
  • Premium Statistics

    • The statistic on this page is a Premium Statistic and is included in this account.

Professional Account

For teams of up to 5 people
$959 USD
per month, billed annually 1
Buy now
  • Free + Premium Statistics
  • Reports
  • Market Insights

1 All prices do not include sales tax. The account requires an annual contract and will renew after one year to the regular list price.

Compare accounts

Statistics on " Generative artificial intelligence (AI) "

The most important statistics
The most important statistics
The most important statistics
The most important statistics
The most important statistics
The most important statistics

Other statistics that may interest you Generative artificial intelligence (AI)

Overview

3

Generative AI tools

3

Generative AI benchmarks

7

Generative AI developers

5

Generative AI use

3

Special focus: United States

5

Further Content: You might find this interesting as well

Topics

Generative artificial intelligence (AI)Artificial intelligence (AI) worldwideArtificial intelligence (AI) use in marketing ChatGPTBard
Access all statistics starting from $2,388 USD yearly *

* For commercial use only

Basic Account

For single users
$0 USD
Always free
Access limited to Free Statistics. Premium Statistics are not included.
  • Free Statistics
Based on your interests

Starter Account

For single users
$199 USD
per month, billed annually 1
Buy now
  • Free Statistics
  • Premium Statistics

    • The statistic on this page is a Premium Statistic and is included in this account.

Professional Account

For teams of up to 5 people
$959 USD
per month, billed annually 1
Buy now
  • Free + Premium Statistics
  • Reports
  • Market Insights

1 All prices do not include sales tax. The account requires an annual contract and will renew after one year to the regular list price.

Compare accounts
Learn more about how Statista can support your business.