Announced in 2016, Gym is an open-source Python library developed to help with the development of support knowing algorithms. It aimed to standardize how environments are defined in AI research study, making released research study more quickly reproducible [24] [144] while providing users with a basic user interface for engaging with these environments. In 2022, new developments of Gym have actually been moved to the library Gymnasium. [145] [146]
Gym Retro

Released in 2018, Gym Retro is a platform for support knowing (RL) research on video games [147] utilizing RL algorithms and research study generalization. Prior RL research focused mainly on enhancing agents to fix single jobs. Gym Retro gives the capability to generalize in between video games with similar principles however various appearances.

RoboSumo

Released in 2017, RoboSumo is a virtual world where humanoid metalearning robot representatives at first do not have understanding of how to even stroll, but are provided the objectives of discovering to move and to press the opposing agent out of the ring. [148] Through this adversarial knowing process, the representatives learn how to adjust to changing conditions. When a representative is then eliminated from this virtual environment and positioned in a brand-new virtual environment with high winds, the agent braces to remain upright, recommending it had learned how to stabilize in a generalized method. [148] [149] OpenAI’s Igor Mordatch argued that competition between representatives might develop an intelligence “arms race” that could increase an agent’s capability to work even outside the context of the competition. [148]
OpenAI 5

OpenAI Five is a team of five OpenAI-curated bots utilized in the competitive five-on-five video game Dota 2, that find out to play against human gamers at a high ability level totally through trial-and-error algorithms. Before becoming a group of 5, the first public demonstration happened at The International 2017, the annual premiere championship competition for the video game, where Dendi, an expert Ukrainian player, lost against a bot in a live one-on-one match. [150] [151] After the match, CTO Greg Brockman explained that the bot had found out by playing against itself for 2 weeks of actual time, and that the learning software application was a step in the instructions of producing software application that can deal with complex tasks like a cosmetic surgeon. [152] [153] The system uses a form of support knowing, as the bots discover in time by playing against themselves hundreds of times a day for months, and are rewarded for actions such as killing an enemy and taking map goals. [154] [155] [156]
By June 2018, the capability of the bots broadened to play together as a full group of 5, and they had the ability to defeat teams of amateur and semi-professional gamers. [157] [154] [158] [159] At The International 2018, OpenAI Five played in 2 exhibit matches against professional players, however wound up losing both games. [160] [161] [162] In April 2019, OpenAI Five defeated OG, the reigning world champs of the video game at the time, 2:0 in a live exhibit match in San Francisco. [163] [164] The bots’ last public appearance came later on that month, where they played in 42,729 overall games in a four-day open online competition, winning 99.4% of those games. [165]
OpenAI 5’s mechanisms in Dota 2’s bot gamer reveals the difficulties of AI systems in multiplayer online battle arena (MOBA) games and how OpenAI Five has demonstrated the usage of deep support knowing (DRL) agents to attain superhuman competence in Dota 2 matches. [166]
Dactyl

Developed in 2018, Dactyl utilizes machine learning to train a Shadow Hand, a human-like robot hand, to control physical items. [167] It finds out completely in simulation using the exact same RL algorithms and training code as OpenAI Five. OpenAI tackled the item orientation problem by using domain randomization, a simulation technique which exposes the learner to a variety of experiences instead of trying to fit to truth. The set-up for Dactyl, aside from having motion tracking cameras, also has RGB cameras to allow the robotic to control an approximate object by seeing it. In 2018, OpenAI showed that the system was able to manipulate a cube and an octagonal prism. [168]
In 2019, OpenAI demonstrated that Dactyl could solve a Rubik’s Cube. The robotic had the ability to solve the puzzle 60% of the time. Objects like the Rubik’s Cube present complex physics that is harder to model. OpenAI did this by improving the robustness of Dactyl to perturbations by utilizing Automatic Domain Randomization (ADR), a simulation approach of producing gradually harder environments. ADR varies from manual domain randomization by not needing a human to specify randomization ranges. [169]
API

In June 2020, OpenAI revealed a multi-purpose API which it said was “for accessing new AI designs developed by OpenAI” to let designers get in touch with it for “any English language AI task”. [170] [171]
Text generation

The company has actually popularized generative pretrained transformers (GPT). [172]
OpenAI’s original GPT model (“GPT-1”)

The initial paper on generative pre-training of a transformer-based language design was written by Alec Radford and his colleagues, and released in preprint on OpenAI’s website on June 11, 2018. [173] It demonstrated how a generative model of language could obtain world understanding and process long-range dependences by pre-training on a diverse corpus with long stretches of contiguous text.

GPT-2

Generative Pre-trained Transformer 2 (“GPT-2”) is a not being watched transformer language model and the follower to OpenAI’s initial GPT design (“GPT-1”). GPT-2 was announced in February 2019, with only restricted demonstrative versions initially released to the general public. The complete variation of GPT-2 was not immediately launched due to concern about possible abuse, consisting of applications for composing phony news. [174] Some professionals revealed uncertainty that GPT-2 postured a considerable risk.

In reaction to GPT-2, the Allen Institute for Artificial Intelligence responded with a tool to detect “neural phony news”. [175] Other scientists, such as Jeremy Howard, cautioned of “the technology to totally fill Twitter, email, and the web up with reasonable-sounding, context-appropriate prose, which would muffle all other speech and be difficult to filter”. [176] In November 2019, OpenAI released the total version of the GPT-2 language model. [177] Several websites host interactive presentations of different circumstances of GPT-2 and other transformer models. [178] [179] [180]
GPT-2’s authors argue unsupervised language models to be general-purpose learners, shown by GPT-2 attaining cutting edge accuracy and perplexity on 7 of 8 zero-shot jobs (i.e. the model was not further trained on any task-specific input-output examples).

The corpus it was trained on, called WebText, contains slightly 40 gigabytes of text from URLs shared in Reddit submissions with a minimum of 3 upvotes. It prevents certain issues encoding vocabulary with word tokens by using byte pair encoding. This permits representing any string of characters by encoding both private characters and multiple-character tokens. [181]
GPT-3

First explained in May 2020, Generative Pre-trained [a] Transformer 3 (GPT-3) is a without supervision transformer language design and the follower to GPT-2. [182] [183] [184] OpenAI specified that the complete variation of GPT-3 contained 175 billion specifications, [184] 2 orders of magnitude larger than the 1.5 billion [185] in the full variation of GPT-2 (although GPT-3 models with as few as 125 million specifications were also trained). [186]
OpenAI mentioned that GPT-3 prospered at certain “meta-learning” tasks and could generalize the purpose of a single input-output pair. The GPT-3 release paper provided examples of translation and cross-linguistic transfer learning in between English and Romanian, and in between English and German. [184]
GPT-3 drastically enhanced benchmark outcomes over GPT-2. OpenAI cautioned that such scaling-up of language models could be approaching or experiencing the essential capability constraints of predictive language designs. [187] Pre-training GPT-3 required several thousand petaflop/s-days [b] of calculate, compared to tens of petaflop/s-days for the complete GPT-2 model. [184] Like its predecessor, [174] the GPT-3 trained model was not immediately launched to the general public for concerns of possible abuse, although OpenAI planned to allow gain access to through a paid cloud API after a two-month complimentary private beta that began in June 2020. [170] [189]
On September 23, 2020, GPT-3 was certified specifically to Microsoft. [190] [191]
Codex

Announced in mid-2021, Codex is a descendant of GPT-3 that has actually in addition been trained on code from 54 million GitHub repositories, [192] [193] and is the AI powering the code autocompletion tool GitHub Copilot. [193] In August 2021, an API was launched in private beta. [194] According to OpenAI, the design can develop working code in over a dozen programs languages, the majority of efficiently in Python. [192]
Several concerns with glitches, style flaws and security vulnerabilities were cited. [195] [196]
GitHub Copilot has been implicated of producing copyrighted code, without any author attribution or license. [197]
OpenAI revealed that they would discontinue support for Codex API on March 23, 2023. [198]
GPT-4

On March 14, 2023, OpenAI revealed the release of Generative Pre-trained Transformer 4 (GPT-4), capable of accepting text or image inputs. [199] They announced that the upgraded innovation passed a simulated law school bar exam with a score around the top 10% of test takers. (By contrast, GPT-3.5 scored around the bottom 10%.) They said that GPT-4 could likewise read, analyze or produce up to 25,000 words of text, and write code in all major programming languages. [200]
Observers reported that the version of ChatGPT using GPT-4 was an enhancement on the previous GPT-3.5-based version, with the caveat that GPT-4 retained some of the issues with earlier revisions. [201] GPT-4 is also capable of taking images as input on ChatGPT. [202] OpenAI has decreased to reveal different technical details and data about GPT-4, such as the exact size of the design. [203]
GPT-4o

On May 13, 2024, OpenAI announced and released GPT-4o, which can process and generate text, images and audio. [204] GPT-4o attained state-of-the-art results in voice, multilingual, and vision criteria, setting brand-new records in audio speech recognition and translation. [205] [206] It scored 88.7% on the Massive Multitask Language Understanding (MMLU) criteria compared to 86.5% by GPT-4. [207]
On July 18, 2024, OpenAI launched GPT-4o mini, a smaller sized variation of GPT-4o changing GPT-3.5 Turbo on the ChatGPT user interface. Its API costs $0.15 per million input tokens and $0.60 per million output tokens, compared to $5 and $15 respectively for GPT-4o. OpenAI anticipates it to be particularly useful for enterprises, startups and developers seeking to automate services with AI representatives. [208]
o1

On September 12, 2024, OpenAI released the o1-preview and o1-mini designs, which have actually been created to take more time to think of their reactions, causing greater accuracy. These models are particularly effective in science, coding, and thinking jobs, and were made available to ChatGPT Plus and Team members. [209] [210] In December 2024, o1-preview was replaced by o1. [211]
o3

On December 20, 2024, OpenAI revealed o3, the successor of the o1 thinking model. OpenAI also revealed o3-mini, a lighter and much faster version of OpenAI o3. As of December 21, 2024, this model is not available for public use. According to OpenAI, they are evaluating o3 and o3-mini. [212] [213] Until January 10, 2025, safety and security scientists had the chance to obtain early access to these designs. [214] The design is called o3 rather than o2 to prevent confusion with telecommunications companies O2. [215]
Deep research study

Deep research study is a representative developed by OpenAI, revealed on February 2, 2025. It leverages the abilities of OpenAI’s o3 model to perform extensive web browsing, data analysis, and synthesis, delivering detailed reports within a timeframe of 5 to thirty minutes. [216] With browsing and Python tools enabled, it reached an accuracy of 26.6 percent on HLE (Humanity’s Last Exam) standard. [120]
Image category

CLIP

Revealed in 2021, CLIP (Contrastive Language-Image Pre-training) is a model that is trained to examine the semantic similarity between text and images. It can significantly be used for image classification. [217]
Text-to-image

DALL-E

Revealed in 2021, DALL-E is a Transformer model that produces images from textual descriptions. [218] DALL-E utilizes a 12-billion-parameter variation of GPT-3 to interpret natural language inputs (such as “a green leather handbag formed like a pentagon” or “an isometric view of a sad capybara”) and create corresponding images. It can develop images of practical things (“a stained-glass window with an image of a blue strawberry”) as well as things that do not exist in truth (“a cube with the texture of a porcupine”). Since March 2021, no API or code is available.

DALL-E 2

In April 2022, OpenAI announced DALL-E 2, an upgraded variation of the model with more reasonable outcomes. [219] In December 2022, OpenAI published on GitHub software for Point-E, a new fundamental system for converting a text description into a 3-dimensional model. [220]
DALL-E 3

In September 2023, OpenAI revealed DALL-E 3, a more powerful design much better able to generate images from intricate descriptions without manual prompt engineering and render complex details like hands and text. [221] It was released to the general public as a ChatGPT Plus feature in October. [222]
Text-to-video

Sora

Sora is a text-to-video design that can create videos based upon brief detailed triggers [223] along with extend existing videos forwards or in reverse in time. [224] It can create videos with resolution up to 1920x1080 or 1080x1920. The optimum length of generated videos is unidentified.

Sora’s development group called it after the Japanese word for “sky”, to symbolize its “limitless innovative potential”. [223] Sora’s innovation is an adjustment of the technology behind the DALL · E 3 text-to-image model. [225] OpenAI trained the system utilizing publicly-available videos along with copyrighted videos accredited for that purpose, however did not expose the number or the precise sources of the videos. [223]
OpenAI demonstrated some Sora-created high-definition videos to the general public on February 15, 2024, mentioning that it could produce videos as much as one minute long. It likewise shared a technical report highlighting the approaches used to train the design, and the model’s abilities. [225] It acknowledged some of its imperfections, consisting of struggles mimicing complicated physics. [226] Will Douglas Heaven of the MIT Technology Review called the presentation videos “impressive”, however noted that they should have been cherry-picked and might not represent Sora’s typical output. [225]
Despite uncertainty from some academic leaders following Sora’s public demo, notable entertainment-industry figures have revealed significant interest in the innovation’s potential. In an interview, actor/filmmaker Tyler Perry revealed his awe at the innovation’s capability to create reasonable video from text descriptions, citing its prospective to change storytelling and material production. He said that his enjoyment about Sora’s possibilities was so strong that he had actually decided to stop briefly prepare for broadening his Atlanta-based film studio. [227]
Speech-to-text

Whisper

Released in 2022, Whisper is a general-purpose speech acknowledgment design. [228] It is trained on a large dataset of diverse audio and is also a multi-task model that can carry out multilingual speech recognition as well as speech translation and language recognition. [229]
Music generation

MuseNet

Released in 2019, pediascape.science MuseNet is a deep neural net trained to predict subsequent musical notes in MIDI music files. It can generate songs with 10 instruments in 15 styles. According to The Verge, a tune generated by MuseNet tends to start fairly however then fall under mayhem the longer it plays. [230] [231] In pop culture, preliminary applications of this tool were utilized as early as 2020 for the web mental thriller Ben Drowned to develop music for the titular character. [232] [233]
Jukebox

Released in 2020, Jukebox is an open-sourced algorithm to create music with vocals. After training on 1.2 million samples, the system accepts a category, artist, and a snippet of lyrics and outputs tune samples. OpenAI specified the songs “reveal local musical coherence [and] follow conventional chord patterns” however acknowledged that the tunes lack “familiar bigger musical structures such as choruses that duplicate” which “there is a considerable gap” in between Jukebox and human-generated music. The Verge mentioned “It’s technically excellent, even if the results sound like mushy variations of songs that might feel familiar”, while Business Insider specified “surprisingly, a few of the resulting songs are memorable and sound legitimate”. [234] [235] [236]
User user interfaces

Debate Game

In 2018, OpenAI released the Debate Game, which teaches devices to debate toy issues in front of a human judge. The is to research whether such a method may help in auditing AI choices and in establishing explainable AI. [237] [238]
Microscope

Released in 2020, Microscope [239] is a collection of visualizations of every substantial layer and nerve cell of 8 neural network designs which are frequently studied in interpretability. [240] Microscope was created to examine the features that form inside these neural networks quickly. The models included are AlexNet, VGG-19, different versions of Inception, and different variations of CLIP Resnet. [241]
ChatGPT

Launched in November 2022, ChatGPT is an expert system tool built on top of GPT-3 that provides a conversational interface that permits users to ask questions in natural language. The system then responds with an answer within seconds.