TellWorldAI World Archive
Ask中文
News
Models
Organisations
People
Papers & concepts
Products & agents
Chips & robots
History
Topics
Ask中文

Home›Models

Model

GPT-4

GPT-4 is OpenAI's large multimodal model, released on 14 March 2023. It accepts image and text input, scored near the top of human test-takers on a simulated bar exam, and was succeeded by GPT-4 Turbo, GPT-4o and GPT-5.

Status DeprecatedModality Text, ImageContext (K tokens) 32 Era Generative AI for everyone

License
Proprietary
Access
API, App
Release date
Developer
OpenAI

What it is

GPT-4 is a large multimodal model from OpenAI, released on 14 March 2023. It accepts image and text inputs and produces text outputs.[1][2] It was offered to paying ChatGPT Plus users and through the API. CNBC reported API pricing of about 3 cents per roughly 750 words of prompt and 6 cents per roughly 750 words of response.[3] The technical report was posted to arXiv on 15 March 2023.[1]

Origins

GPT-4 is a successor to GPT-3.5.[5] Jakub Pachocki, OpenAI's research director from 2021, led its development.[6] Training finished in August 2022, and from that month OpenAI recruited more than 50 external experts to test the model adversarially.[1]

Main features

The technical report describes a Transformer-style model pre-trained to predict the next token, using publicly available data and data licensed from third parties. It was then fine-tuned with reinforcement learning from human feedback (RLHF).[1]

OpenAI did not disclose the architecture, including model size, the hardware and training compute, how the dataset was built, or the training method. It cited the competitive landscape and safety implications as reasons.[1]

On a simulated Uniform Bar Examination, GPT-4 scored around the top 10% of test takers, while GPT-3.5 scored around the bottom 10%.[1] It reached 86.4% on the MMLU benchmark. On translated versions of MMLU it surpassed the previous English-language state of the art in 24 of the 26 languages tested.[1]

OpenAI said it could predict some aspects of GPT-4's performance from smaller models trained with no more than 1/1,000th of its compute.[1]

Compared with GPT-3.5, OpenAI reported that GPT-4's tendency to respond to requests for disallowed content fell by 82%, and that it handled sensitive requests in line with its policies 29% more often.[1]

Its place in development

Microsoft said its Bing AI chatbot used GPT-4, and OpenAI said the model was trained on Microsoft Azure.[3] GPT-4 was followed by GPT-4 Turbo in November 2023, GPT-4o in May 2024 and GPT-5 in August 2025.[5]

The older snapshots have been retired on a schedule. The gpt-4-32k snapshots were shut down on 6 June 2025, and gpt-4-0314 on 26 March 2026. The gpt-4-0613 snapshot was scheduled for shutdown on 23 October 2026.[4]

CapabilitiesLarge language modelsMultimodal

TechniquesFoundation models

Governance & standardsSafety & governance

Version history

22
  1. 2018
    • Jun 11, 2018 GPT-1

      GPT-1 was OpenAI's first generative pre-trained transformer, published in June 2018. It showed that one language model, pre-trained on unlabelled text and then fine-tuned, could improve results on many language tasks.

  2. 2019
    • Feb 14, 2019 GPT-2

      GPT-2 is a large language model that OpenAI announced in February 2019. Its largest version has 1.5 billion parameters. OpenAI first withheld the full model, then released it in stages through November 2019.

  3. 2020
    • May 28, 2020 GPT-3

      GPT-3 is an autoregressive language model with 175 billion parameters, described in a May 2020 paper. It could perform new tasks from a few examples in the prompt, without any gradient updates, and it was first offered through OpenAI's API.

  4. 2022
    • Mar 15, 2022 GPT-3.5 (text-davinci-002/003)

      GPT-3.5 is the family of OpenAI language models between GPT-3 and GPT-4, released through API models such as text-davinci-002 and text-davinci-003 and underlying the first version of ChatGPT.

  5. 2023
    • Mar 1, 2023 GPT-3.5 Turbo

      GPT-3.5 Turbo was the chat-optimised GPT-3.5 model OpenAI released through its API on 1 March 2023, priced at a tenth of the cost of earlier models and later replaced in ChatGPT by GPT-4o mini.

    • Mar 14, 2023 GPT-4 · This page

      GPT-4 is OpenAI's large multimodal model, released on 14 March 2023. It accepts image and text input, scored near the top of human test-takers on a simulated bar exam, and was succeeded by GPT-4 Turbo, GPT-4o and GPT-5.

    • Nov 6, 2023 GPT-4 Turbo

      GPT-4 Turbo is a cheaper, more capable version of OpenAI's GPT-4, announced on 6 November 2023 with a 128,000-token context window and lower API prices.

  6. 2024
    • May 13, 2024 GPT-4o

      GPT-4o is a multimodal model from OpenAI, released in May 2024, that takes text, audio, image and video as input and generates text, audio and image output within one neural network. It was removed from ChatGPT in February 2026.

    • Jul 18, 2024 GPT-4o mini

      A small, low-cost language model OpenAI released on 18 July 2024, priced far below GPT-4o and reported at 82% on MMLU; it replaced GPT-3.5 Turbo in ChatGPT on launch day.

    • Sep 12, 2024 OpenAI o1

      OpenAI o1 is the first reasoning model series from OpenAI, introduced in September 2024. It thinks through a long internal chain of thought before answering, and it was the base for later reasoning models such as o3.

  7. 2025
    • Feb 27, 2025 GPT-4.5

      GPT-4.5 is a large language model OpenAI released on 27 February 2025 under the code name "Orion". It was its largest chat model at the time, priced far above GPT-4o, and was retired from general availability within months.

    • Apr 14, 2025 GPT-4.1 (with mini and nano)

      OpenAI's April 2025 family of three API models — GPT-4.1, GPT-4.1 mini and GPT-4.1 nano — offering a context window of over a million tokens, low prices per token and stronger coding scores. It was succeeded by GPT-5.

    • Apr 16, 2025 OpenAI o3

      OpenAI's second-generation reasoning model, announced in December 2024 and released in April 2025, which posted large gains over o1 on science, software-engineering and competitive-programming benchmarks before being succeeded by GPT-5.

    • Aug 7, 2025 GPT-5

      GPT-5 is OpenAI's fifth GPT foundation model, released on 7 August 2025. It pairs a fast model with a deeper reasoning model and a router that chooses between them for each request.

    • Nov 12, 2025 GPT-5.1

      GPT-5.1 is an update to GPT-5 released by OpenAI on 12 November 2025. It added a warmer default personality with selectable styles, and its API version has been scheduled for shutdown in 2027.

    • Dec 11, 2025 GPT-5.2

      GPT-5.2 is a model family OpenAI released on 11 December 2025, with Instant, Thinking and Pro versions and a coding variant. It was aimed at professional work such as spreadsheets and presentations.

  8. 2026
    • Mar 5, 2026 GPT-5.4

      GPT-5.4 is a model OpenAI released on 5 March 2026 in Thinking and Pro versions. It added built-in computer use and improved deep research, and it was the model that succeeded GPT-5.2.

    • Apr 23, 2026 GPT-5.5

      GPT-5.5 is an OpenAI model released on 23 April 2026 as GPT-5.5 Thinking and Pro, with an API version the next day. It succeeded GPT-5.4 and had a cybersecurity variant in limited preview.

    • Jul 8, 2026 GPT-Live-1

      GPT-Live-1 is OpenAI's full-duplex voice model family, launched 8 July 2026, which lets ChatGPT listen and speak at the same time and replaced the earlier chained speech pipeline in ChatGPT Voice.

    • Jul 9, 2026 GPT-5.6 (Sol, Terra and Luna)

      OpenAI's three-model family released on 9 July 2026 — Luna, Terra and Sol, ordered from least to most capable — with Sol pitched as the company's strongest coding and agent model.

    • Sep 3, 2026 GPT-6 (Astra, Sol, Luna, Instant)

      GPT-6 is OpenAI's sixth-generation model family, launched from 3 September 2026 with GPT-6 Astra, followed by Sol, Luna and Instant. Astra was the first OpenAI model to reach its internal critical cybersecurity threshold.

    • Sep 29, 2026 GPT-6.1 (Sol and Terra)

      OpenAI's late-2026 model family: Sol, released at DevDay on 29 September 2026, and the faster, cheaper Terra on 26 October 2026. A planned Astra version was withheld over safety concerns.

20241

  1. ReleaseOpenAI introduces GPT-4o

    On 13 May 2024 OpenAI announced GPT-4o ("GPT-4 Omni"), its new flagship model, which it said can reason across audio, vision and text in real time. Developers could use it through the API the same day.

20234

  1. UpdateBing Chat is renamed Microsoft Copilot

    On 15 November 2023 Microsoft renamed Bing Chat and Bing Chat Enterprise to Copilot, giving the assistant its own site and a commercial version with data protection.

  2. ReleaseMicrosoft introduces Microsoft 365 Copilot

    On 16 March 2023 Microsoft introduced Microsoft 365 Copilot, pairing GPT-4 and other large language models with business data, and made it generally available in November.

  3. ReleaseOpenAI introduces GPT-4

    On 14 March 2023 OpenAI announced GPT-4, a multimodal model that accepts image and text input and produces text output. OpenAI reported human-level results on several professional and academic benchmarks but disclosed little about the model's design.

  4. ReleaseMicrosoft launches the AI-powered Bing with Bing Chat

    On 7 February 2023 Microsoft unveiled an AI-powered Bing and Edge, built on a new OpenAI model, whose chat feature was launched as a waitlisted preview called Bing Chat.

Developed by1

OpenAI

Builds on1

Transformer

Foundation for1

GPT-4V (GPT-4 with vision)

Succeeded by2

GPT-5GPT-4 Turbo

Successor of2

GPT-3GPT-3.5 (text-davinci-002/003)

Powers4

ChatGPTOpenAI API platformMicrosoft CopilotMicrosoft 365 Copilot

Contributors1

Jakub Pachocki

Uses2

Reinforcement learning from human feedback (RLHF)Transformer

Notes and sources

  1. GPT-4 Technical Reporthttps://arxiv.org/abs/2303.08774
  2. GPT-4https://openai.com/index/gpt-4-research/
  3. OpenAI announces GPT-4, says beats 90% of humans on SAThttps://www.cnbc.com/2023/03/14/openai-announces-gpt-4-says-beats-90percent-of-humans-on-sat.html
  4. Deprecationshttps://developers.openai.com/api/docs/deprecations
  5. OpenAIhttps://en.wikipedia.org/wiki/OpenAI
  6. Jakub Pachockihttps://en.wikipedia.org/wiki/Jakub_Pachocki

Cite this page: GPT-4. TellWorld, https://www.tellworld.net/models/gpt4. Updated 2026-11-08