OpenAI Models Reference

Complete capabilities matrix for currently documented OpenAI API models (non-deprecated)

Last updated: 2025-11-19 | Source: OpenAI Platform Documentation

Legend

✅ = Supported ❌ = Not supported Ⓡ = Reasoning model family

Vision In (OCR): Model can READ/analyze images (for OCR, image understanding)
Vision Out (Gen): Model can GENERATE images/videos

General-Purpose GPT + Reasoning Models

Main LLMs for chat completions, responses, and agent tools. These are the "brain" models you'd typically use.

Model ID	Type	Vision In (OCR)	Vision Out (Gen)	Audio In	Audio Out	Chat/Text	Tools	Reasoning
`gpt-5.1`	Flagship GPT	✅	❌	❌	❌	✅	✅	✅ Configurable
`gpt-5`	GPT Family	✅	❌	❌	❌	✅	✅	✅ Oriented
`gpt-5-pro`	Reasoning GPT	✅	❌	❌	❌	✅	✅	✅ Advanced
`gpt-5-mini`	Smaller GPT-5	✅	❌	❌	❌	✅	✅	✅ Lighter
`gpt-5.1-codex`	Coding-Optimized	✅	❌	❌	❌	✅	✅ Strong	✅ Coding
`gpt-5.1-codex-mini`	Small Codex	✅	❌	❌	❌	✅	✅	✅ Lighter
`gpt-4.1`	High-Intelligence	✅	❌	❌	❌	✅	✅ Excellent	❌
`gpt-4.1-mini`	Small, Fast	✅	❌	❌	❌	✅	✅ Optimized	❌
`gpt-4.1-nano`	Tiny GPT-4.1	✅	❌	❌	❌	✅	✅	❌
`gpt-4o`	Omni GPT	✅	❌	✅ Native	✅ Native	✅	✅	❌
`gpt-4o-mini`	Smaller Omni	✅	❌	✅ Variants	✅ TTS	✅	✅	❌
`gpt-4`	Legacy GPT-4	✅	❌	❌	❌	✅	✅	❌
`gpt-4-turbo-preview`	Legacy Turbo	✅	❌	❌	❌	✅	✅	❌
`o3`	Ⓡ Reasoning	✅ Strong	❌	❌	❌	✅	✅	✅ Slow/Strong
`o3-mini`	Ⓡ Fast	✅	❌	❌	❌	✅	✅	✅ Cheaper
`o4-mini`	Ⓡ Reasoning	✅	❌	❌	❌	✅	✅	✅
`o3-deep-research`	Ⓡ Web Research	✅	❌	❌	❌	✅ Reports	✅ Web Search	✅ Multi-Step
`o4-mini-deep-research`	Ⓡ Fast Research	✅	❌	❌	❌	✅	✅	✅
`gpt-oss`	Small OSS-style	❌ Text Only	❌	❌	❌	✅	✅ Standard	❌
`davinci-002`	Legacy Base	❌	❌	❌	❌	✅	✅ Basic	❌

All GPT and reasoning models support structured JSON output and function/tool calling via standard APIs. Many have multiple snapshots (dated model IDs) and aliases.

Embedding & Moderation Models

Models for generating vector embeddings and content moderation.

Model ID	Type	Vision In (OCR)	Embedding	Description
`text-embedding-3-large`	Embeddings	❌	✅	High-quality, multilingual
`text-embedding-3-small`	Embeddings	❌	✅	Cheap, fast
`text-embedding-ada-002`	Legacy	❌	✅	Older, backward-compatible
`omni-moderation-latest`	Moderation	✅ Image	❌	Text + image moderation
`text-moderation-latest`	Moderation	❌	❌	Text-only moderation

Audio & Speech Models

Models for speech-to-text, text-to-speech, and audio-capable chat via the Audio API.

Model ID	Type	Vision In (OCR)	Audio In	Audio Out	Chat	Tools
`gpt-audio-mini`	General Audio LLM	❌	✅	✅	✅	✅
`gpt-4o-audio-preview`	4o Audio	✅	✅	✅	✅	✅
`gpt-4o-mini-audio-preview`	4o Mini Audio	✅	✅	✅	✅	✅
`gpt-4o-mini-tts`	TTS	❌	❌	✅ 11+ Voices	❌	❌
`gpt-4o-transcribe`	STT	❌	✅	❌	Outputs text	❌
`gpt-4o-mini-transcribe`	Small STT	❌	✅	❌	Outputs text	❌
`gpt-4o-transcribe-diarize`	STT + Diarization	❌	✅	❌	Text + speakers	❌
`whisper-1`	Classic STT	❌	✅	❌	Text only	❌

Image & Vision-Focused Models

Models for image generation and manipulation via the Images API. These models GENERATE images (Vision Out), unlike chat models that READ images (Vision In).

Model ID	Type	Vision In (OCR)	Vision Out (Gen)	Description
`gpt-image-1`	Image Gen LMM	✅ Text + Image	✅ Generate	Multimodal LLM via Images API
`dall-e-3`	Image Gen	❌	✅ Generate	Older image generation
`dall-e-2`	Legacy Image Gen	❌	✅ Generate	Legacy image generation

gpt-image-1 is a multimodal LLM that can both READ images (Vision In) and GENERATE images (Vision Out). You call it via the Images API / image-tool, not as a general chat model.

Realtime Models

Streaming text + audio models via WebRTC/WebSocket with API key authentication.

Model ID	Type	Vision In (OCR)	Audio In	Audio Out	Chat	Tools
`gpt-realtime-2025-08-28`	Realtime GPT	Check docs	✅	✅	✅ Streaming	✅ Realtime API

API Access

All models listed above are accessible via server-side code using an OpenAI API key.

API Endpoints

chat or responses - GPT + reasoning + audio
embeddings - Vector embeddings
audio - STT/TTS
images - Image generation
moderations - Content moderation
realtime - WebRTC/WebSocket streaming

Additional Info

Models listed are non-deprecated as of 2025-11-19
Many have snapshot IDs and aliases
Some eval-only graders not listed for brevity
Check live docs for updates

Recommended Models for George AI

💬 Chat Assistants

gpt-4o-mini - Fast, cheap
gpt-4.1 - High intelligence
gpt-5.1 - Flagship with reasoning

🔍 Embeddings

text-embedding-3-small - Fast, cheap
text-embedding-3-large - High quality

👁️ Vision In (OCR)

Reading/analyzing images

gpt-4o - Native multimodal
gpt-4.1 - Best quality
gpt-5.1 - Latest

🎨 Vision Out (Gen)

Generating images

gpt-image-1 - Latest (LMM)
dall-e-3 - High quality

✨ Structured Data Extraction

Recommended: gpt-4o-mini (fast/cheap), gpt-4.1-mini (optimized for tools), o3-mini (reasoning)

💡 Tip: All OpenAI GPT models support native function calling. Use reasoning models (o-series) for complex extraction tasks requiring multi-step logic.