OpenAI GPT-4 at a glance
GPT-4 accepted text and image inputs, scored around the top 10% on a simulated bar exam, and launched in ChatGPT Plus and through an API waitlist.
What changed with OpenAI GPT-4
On March 14, 2023, OpenAI announced GPT-4, a large multimodal model that takes text and images as input and returns text. OpenAI reported that it scored around the top 10% of test takers on a simulated Uniform Bar Exam, where GPT-3.5 had landed near the bottom 10%, and said six months of alignment work had produced its best results yet on factuality and steerability.
Text input launched in ChatGPT Plus and in the API through a waitlist, while image input stayed with a single partner at first. The API offered an 8,192-token model at $0.03 per 1K prompt tokens and $0.06 per 1K completion tokens, plus limited access to a 32,768-token version. OpenAI also open-sourced its Evals framework and warned that GPT-4 could still make up facts.
OpenAI GPT-4 benchmarks and pricing
Values are copied from the linked source on the date shown. They are not IA Compa measurements.
- Input price per 1M tokensas of · Source for Input price per 1M tokens: openai.com (opens in a new tab)
- $30 ($0.03 per 1K, 8K model)
- Output price per 1M tokensas of · Source for Output price per 1M tokens: openai.com (opens in a new tab)
- $60 ($0.06 per 1K, 8K model)
- Context windowas of · Source for Context window: openai.com (opens in a new tab)
- 8,192 tokens (32K in limited access)
How it compares
Other models in the same category that we have covered, using each model's latest recorded value. (Models)
Where to compare OpenAI GPT-4
Independent leaderboards and official resources for comparing this model yourself. Rankings change often; each site shows its own current data.
General leaderboards
Official resources
Why OpenAI GPT-4 matters for businesses
At launch, GPT-4 set the bar that other labs chased for about a year. For businesses it made drafting, analysis and coding help reliable enough for real work, while its price kept it reserved for higher-value tasks, with human review still required.
Sources for this story
- GPT-4 (opens in a new tab)Official
OpenAI · openai.com
- GPT-4 (archived March 20, 2023) (opens in a new tab)Official
OpenAI via Internet Archive · web.archive.org
- GPT-4 model page (opens in a new tab)Benchmark
Artificial Analysis · artificialanalysis.ai
Last verified: . IA Compa summarizes public announcements for business readers. We are not affiliated with the companies mentioned. Always confirm details, pricing, and availability with the official source.
- #llm
- #openai
- #gpt-4
- #multimodal
- #frontier-model
