AI provider / OpenAI
OpenAI
A provider commonly associated with conversational AI, multimodal assistants, reasoning-oriented workflows and developer-facing model platforms.
OpenAI is the frontier lab that turned a research bet on generative pre-training into ChatGPT — the product that reset public expectations of AI almost overnight. From GPT-1 in 2018 to the GPT-5 series, its catalog has become the default reference point for general-purpose assistants, coding and multimodal interaction.
Its line now runs on three tracks: the general-purpose GPT family, the reasoning-focused o-series built around test-time compute, and the open-weight gpt-oss branch. Shipped at a relentless cadence and wired directly into a developer platform, that mix keeps OpenAI at the center of how teams build with AI.
Model catalog
Every release in this provider’s model line.
Each row is a distinct model release — its date, its type and the traits it is known for. Open a source to dig into any of them.
| Model | Released | Type | Key traits | Source |
|---|---|---|---|---|
| GPT-5.5 | Apr 2026 | Flagship |
| Wikipedia |
| GPT-5.4 | Mar 2026 | Version update |
| Wikipedia |
| GPT-5.2 | Dec 2025 | Version update |
| GPT version timeline |
| GPT-5.1 | Nov 2025 | Version update |
| OpenAI release notes |
| GPT-5 | Aug 2025 | Flagship |
| OpenAI |
| gpt-oss | Aug 2025 | Open release |
| OpenAI |
| o3 & o4-mini | Apr 2025 | Reasoning |
| OpenAI |
| GPT-4.5 | Feb 2025 | Flagship |
| OpenAI |
| o1-preview | Sep 2024 | Reasoning |
| OpenAI |
| GPT-4o | May 2024 | Multimodal |
| OpenAI |
| GPT-4 Turbo | Nov 2023 | Version update |
| OpenAI DevDay |
| GPT-4 | Mar 2023 | Flagship |
| OpenAI |
| ChatGPT (GPT-3.5) | Nov 2022 | Product launch |
| OpenAI |
| GPT-3 | Jun 2020 | Scale-up |
| OpenAI |
| GPT-2 | Feb 2019 | Version update |
| OpenAI |
| GPT-1 | Jun 2018 | Research |
| OpenAI |
Timeline
A practical timeline of model releases and company milestones.
This page tracks the moments that matter most: major model releases, version jumps and company moves that changed how this provider shows up in the AI market.
GPT-5.5 ships as OpenAI’s smartest model to date.
The current top of the GPT line, positioned on frontier reasoning and agentic reliability rather than raw novelty.
GPT-5.4 lands, followed by mini and nano tiers within two weeks.
The cadence itself becomes the story: flagship updates now arrive quarterly with size tiers trailing days behind.
GPT-5.2 sharpens coding and long-context agent work.
A quick follow-up release aimed squarely at competitive pressure on developer and agent workloads at the end of 2025.
GPT-5.1 tunes the flagship for warmth and instruction following.
Instant and Thinking variants ship together, with adaptive reasoning effort becoming the main lever for quality versus speed.
GPT-5 unifies fast responses and deep reasoning in one system.
A router decides when to think hard, folding the GPT and o-series identities into a single flagship with sizes for every tier.
gpt-oss returns OpenAI to open weights after GPT-2.
Two Apache-licensed reasoning models, 120B and 20B, give the company an answer to the open-weight ecosystem it had left to others.
o3 and o4-mini bring agentic tool use to the reasoning series.
The o-series learns to chain web search, code execution and image reasoning inside its thinking, previewing the agent direction of later flagships.
GPT-4.5 ships as the largest pure pre-training push of the GPT line.
Positioned around better world knowledge and fewer hallucinations, it marks the last big non-reasoning step before unification under GPT-5.
o1-preview opens the o-series and the test-time-compute era.
Models that think before answering trade latency for accuracy on math, science and code, starting a second product line beside GPT.
GPT-4o makes the flagship natively multimodal in real time.
One model handles text, vision and audio with conversational latency, and the omni design becomes the new default for assistants.
GPT-4 Turbo brings 128K context and cheaper tokens at DevDay.
OpenAI pairs a faster, cheaper flagship with GPTs and the Assistants API, pushing the platform story as hard as the model itself.
GPT-4 raises the bar on reasoning and adds image understanding.
The first GPT generation to pass professional exams convincingly becomes the benchmark every other frontier model is measured against.
ChatGPT launches on GPT-3.5 and resets public expectations of AI.
A chat interface over an instruction-tuned model reaches a hundred million users in two months and makes conversational AI mainstream.
GPT-3 arrives at 175B parameters behind the first OpenAI API.
Few-shot prompting at scale turns language models into a product surface, and the API becomes the template for the whole industry.
GPT-2 scales to 1.5B parameters and sparks the release debate.
A staged release driven by misuse concerns makes GPT-2 the first model whose rollout itself became an AI-safety story.
GPT-1 introduces generative pre-training for language.
OpenAI shows that pre-training a transformer on raw text and fine-tuning it downstream beats task-specific architectures, starting the GPT line.