GPT-4 arrives in ChatGPT Plus with API access by waitlist

OpenAI announced GPT-4 on 14 March 2023, opening text-based access through ChatGPT Plus and a developer API waiting list.

Plus subscribers receive access with a usage cap that OpenAI says will depend on demand and system performance. API invitations begin today and will expand gradually.

GPT-4 is designed to take text and images as inputs and produce text. Image input remains a research preview and is not publicly available, so the initial text release should not be mistaken for a general image-analysis service.

A limited accessibility pilot

Be My Eyes announced Virtual Volunteer, a GPT-4-powered feature for blind and low-vision users, on the same day. It is in a closed beta with a small group, while others can register for a waiting list in the app.

The planned feature lets a user send an image and ask questions about it. Be My Eyes says it will offer a connection to a sighted volunteer when the tool cannot answer. Wider availability is a goal for the coming months, not an immediate public launch.

Review remains important

OpenAI’s launch notes describe testing and safeguards around the release, alongside remaining limitations. GPT-4 can still produce convincing but inaccurate information. Its responses need checking, especially where errors could have serious consequences.

Illustration: Sulat. Logo: OpenAI / AREA 17.