OpenAI GPT family (GPT-5 / GPT-4o / o3 / o4)
https://openai.com/index/introducing-gpt-5/
OpenAI's foundation model family — GPT-5 (2025-08 flagship), GPT-4o (multimodal workhorse), o3/o4 reasoning series. Substrate for ChatGPT, the OpenAI API, Microsoft Copilot, Azure OpenAI, and many downstream agents in this catalog (Devin, Codex CLI, Operator, Cursor mode). Parametric memory in weights; discrete user memory ships in ChatGPT Memory feature (separate row).
At a glance
- Type
- Frontier foundation model family (GPT-5 / GPT-4o / o3 / o4)
- Tier
- T1
- Created
- 2018 (GPT-1); 2020-06 (GPT-3); 2023-03 (GPT-4); 2024-05 (GPT-4o); 2024-09 (o1); 2025-01 (o3); 2025-08 (GPT-5)
- Latest release
- GPT-5 (2025-08), o4-mini, GPT-4.1 series
- License
- not applicable — not OSS
- GitHub
- not applicable — no GitHub repo
- Pricing
- Pay-per-token API (GPT-5 $1.25/$10 per 1M; GPT-4o $2.50/$10; o3 $2/$8); ChatGPT Free + Plus ($20/mo) + Pro ($200/mo) + Team + Enterprise
- Funding
- $57.9B+ total raised; $500B valuation (2025 secondary tender); Microsoft $13B+ strategic investment
Taxonomy
- storage
- parametric
- retrieval
- parametric-recall
- persistence
- parametric-permanent
- update
- read-only
- unit
- weight
- governance
- opaque
- conflict
- n/a
When to use
Optimised for: general-purpose reasoning, coding, multimodal, voice/real-time
Anti-fit: weights closed; data residency limited (US/EU/Asia regions only); rate-limited without enterprise; no open-weights option; Sora video gated
Pros & cons
Pros
Most-deployed frontier LLM (800M+ MAU ChatGPT); full multimodal stack (vision + audio + image gen + video gen); largest ecosystem of fine-tunes / RAG / agent integrations; Azure distribution.
Cons
Closed weights; expensive at Pro/o3 tier; legal exposure (multiple ongoing NYT / Authors Guild lawsuits); Sora & advanced features behind waitlists; SWE-bench trails Claude Opus 4.5.
Claims & capabilities
GPT-5 Aug-2025 unified reasoning + non-reasoning; o3 first to score >85% on AIME 2025; GPT-4o multimodal (vision/audio/text); 128k context (GPT-5: 400k); function-calling, structured outputs, real-time API
Technical surface
- API surface
- REST + SDK (Python, TS, Java, Go, .NET); Azure OpenAI variant; Realtime API (WebRTC)
- Backend storage
- not applicable — substrate foundation model
- Deployment
- Managed-only (OpenAI API + Microsoft Azure OpenAI)
- Embedding model
- not applicable — not a memory product
- Multi-tenancy
- not applicable — substrate foundation model
- MCP
- not applicable — substrate foundation model
- A2A
- not applicable — substrate foundation model
- OpenTelemetry
- not applicable — substrate foundation model
Compare OpenAI GPT family (GPT-5 / GPT-4o / o3 / o4) with…
- OpenAI GPT family (GPT-5 / GPT-4o / o3 / o4) vs Anthropic Claude (foundation models)
- OpenAI GPT family (GPT-5 / GPT-4o / o3 / o4) vs Google Gemini 3 family
- OpenAI GPT family (GPT-5 / GPT-4o / o3 / o4) vs Meta Llama 4 family
- OpenAI GPT family (GPT-5 / GPT-4o / o3 / o4) vs Mistral Large 2 / Mixtral family
Similar systems
Other foundation models (substrate reference) in the catalog, ranked by inbound references.
- Anthropic Claude (foundation models) T1
Claude family of frontier foundation models — Sonnet 4 (workhorse), Opus 4.5 (frontier capability), Haiku 4.5 (fast/cheap). Frontier model used as substrate by Claude Code, Anthropic Memory tool, Cline, Aider, Continue.dev, Cursor, Goose, and dozens of other catalog entries. Memory in Claude is parametric (in weights) — discrete episodic memory lives outside the model in the Claude Memory tool (see Platform-provider memory section).
- Google Gemini 3 family T1
Google DeepMind's Gemini 3 frontier model family — Gemini 3 Pro (top tier), 3 Flash (fast), 3 Nano (on-device). Substrate for Google Workspace AI, Vertex AI Agent Platform, Gemini Code Assist, Project Mariner, and Project Astra. Multimodal-native; 1M-10M context window. Largest context window of any frontier model. Used by Gemini Enterprise Agent Platform Memory Bank (separate row).
- Mistral Large 2 / Mixtral family T1
French frontier-model lab (Paris). Family includes Mistral Large 2 (123B dense, Jul-2024), Mixtral 8x22B (open-weights MoE), Mistral Small 3 (24B, open-weights Jan-2025), Codestral (code specialist). EU-headquartered alternative to US/China model labs; GDPR-friendly data residency. Distinct from the Mistral Le Chat memory feature (already in catalog) — this row covers the model family itself as substrate.
- Meta Llama 4 family T1
Meta's Llama 4 family (released 2025-04) — Scout (109B MoE, 17B active), Maverick (400B MoE, 17B active), Behemoth (2T MoE, 288B active; preview). First Llama generation to use mixture-of-experts and to be multimodal-native. Substrate for many OSS agents (Open-Interpreter, Ollama-served local agents, vLLM-served enterprise inference). Llama community license (commercial use permitted under 700M MAU threshold).
- xAI Grok 4 T1
Elon Musk's xAI frontier model — Grok 4 launched 2025-07. Trained on Colossus (100k+ GPU supercluster in Memphis). Deeply integrated with X/Twitter (real-time tweet data access). Grok 4 Heavy variant for premium tier. No frontier-tier open-weights — Grok 1 (314B MoE) is the only weight release.
- 01.AI Yi family T2
Kai-Fu Lee's 01.AI (founded 2023, Beijing). Yi family open-weights — Yi-34B (2023), Yi-Large (proprietary 2024), Yi-Lightning (2024-10, frontier-tier, matched GPT-4o on LMSYS chatbot arena). Apache 2.0 (Yi-34B and base sizes). Important second Chinese open-weights option alongside DeepSeek + Qwen.
Related systems
Referenced by (49)
- 1X Technologies depends on at runtime — adjacent-infrastructure cell: OpenAI Startup Fund portfolio; Norwegian / US robotics ecosystem
- Adept ACT depends on at runtime — adjacent-infrastructure cell: Amazon AGI (parent post-acquisition); spiritual predecessor to Anthropic Computer Use + OpenAI Operator + Project Mariner
- Agency Swarm depends on at runtime — Agent framework built on OpenAI Assistants API — agencies (roles + comm flow); replaced after Assistants v
- Backboard depends on at runtime — adjacent-infrastructure cell: BYO LLM (OpenAI/Anthropic/Gemini/Ollama keys)
- Browser Use depends on at runtime — adjacent-infrastructure cell: Playwright; Anthropic / OpenAI / Gemini APIs; Browserbase / Hyperbrowser substrates
- ChatGPT — Codex (cloud agent) depends on at runtime — adjacent-infrastructure cell: OpenAI Codex (model) + cloud sandbox
- Cline (harness) depends on at runtime — VS Code extensions; multi-model via standard OpenAI-compatible APIs.
- Cognition Devin v2 / Spec Mode depends on at runtime — adjacent-infrastructure cell: absorbed Windsurf IP (Aug 2025); Anthropic + OpenAI models
- Continue.dev (harness) depends on at runtime — adjacent-infrastructure cell: any OpenAI-compatible API; MCP
- Cursor (IDE) depends on at runtime — adjacent-infrastructure cell: Anthropic Claude + OpenAI GPT + own Cursor-Tab model
- Decagon depends on at runtime — Backed by OpenAI. ElevenLabs partnership for voice (2025).
- Figure AI depends on at runtime — adjacent-infrastructure cell: BMW manufacturing; OpenAI partnership (ended Feb 2025); Microsoft Azure
- GitHub Copilot (Agent Mode) depends on at runtime — adjacent-infrastructure cell: Azure OpenAI + Anthropic Claude + Google Gemini (multi-model 2024+)
- GitHub Copilot Workspace depends on at runtime — adjacent-infrastructure cell: Azure OpenAI + Anthropic Claude (GitHub-hosted models)
- Highlight AI depends on at runtime — adjacent-infrastructure cell: Anthropic / OpenAI APIs; local Llama option
- Hume EVI depends on at runtime — adjacent-infrastructure cell: Twilio (telephony); Anthropic / OpenAI LLMs (or Hume's eLLM)
- JetBrains AI Assistant depends on at runtime — adjacent-infrastructure cell: JetBrains Cloud (OpenAI + Anthropic + Google)
- Letta / MemGPT depends on at runtime — adjacent-infrastructure cell: BYO LLM (OpenAI/Anthropic/local); bundled storage (Postgres/SQLite)
- LiteLLM depends on at runtime — ~14k stars; 100+ LLM providers behind OpenAI-compatible API; widely used as proxy/gateway.
- LiveKit Agents depends on at runtime — adjacent-infrastructure cell: OpenAI Realtime; Anthropic; Deepgram / Cartesia / Eleven STT-TTS; Twilio (for PSTN)
- Lovable.dev — harness depends on at runtime — adjacent-infrastructure cell: Supabase (default DB); Anthropic Claude + OpenAI
- Lutra AI depends on at runtime — adjacent-infrastructure cell: OAuth providers; Anthropic / OpenAI; Google Workspace
- Magnitude depends on at runtime — adjacent-infrastructure cell: Playwright; Anthropic / OpenAI APIs
- Manus AI depends on at runtime — adjacent-infrastructure cell: Anthropic Claude (reported) + OpenAI
- Meilisearch depends on at runtime — eneration + caching via OpenAI / HuggingFace / Ollama. Multi-modal (text + images); hybrid
- MetaGPT (harness) depends on at runtime — adjacent-infrastructure cell: OpenAI + Anthropic + Azure OpenAI
- MultiOn depends on at runtime — adjacent-infrastructure cell: Playwright; Chromium-headless infrastructure; computer-use models (Claude / GPT-5 / Gemini)
- OpenAI Agents SDK depends on at runtime — ~9k★; built on the OpenAI Responses API and tools.
- OpenAI Agents SDK Memory depends on at runtime — adjacent-infrastructure cell: requires OpenAI API; BYO persistence layer
- OpenAI ChatGPT Memory depends on at runtime — backend-storage cell: custom (OpenAI-managed)
- OpenAI Codex CLI depends on at runtime — adjacent-infrastructure cell: OpenAI API (Codex / o-series); MCP for tool extensibility; runs locally with TLS-only network egress
- OpenAI Operator depends on at runtime — backend-storage cell: OpenAI-hosted browser session state; no durable cross-session memory by default
- Opencode (SST) depends on at runtime — adjacent-infrastructure cell: any OpenAI-compatible API; MCP
- OpenHands (All Hands AI) depends on at runtime — adjacent-infrastructure cell: Docker; Anthropic + OpenAI APIs
- Phidata / Agno depends on at runtime — adjacent-infrastructure cell: HuggingFace Transformers; OpenAI / Anthropic / Gemini SDKs; YC W22 alumni network
- Pipecat (Daily) depends on at runtime — adjacent-infrastructure cell: Daily WebRTC; OpenAI Realtime; Anthropic; Deepgram + Cartesia STT/TTS
- RepairAgent depends on at runtime — ~$0.14 USD) per bug using GPT-3.5 pricing
- Retell AI depends on at runtime — adjacent-infrastructure cell: Twilio SIP; in-house STT/TTS; Anthropic / OpenAI LLMs
- Roo Code depends on at runtime — adjacent-infrastructure cell: OpenAI-compatible APIs
- Skyvern depends on at runtime — adjacent-infrastructure cell: Playwright; Postgres; Anthropic/OpenAI/Gemini APIs
- smol-developer depends on at runtime — adjacent-infrastructure cell: OpenAI API (GPT-3.5/4); single-file Python script; no other dependencies beyond LLM provider
- Sourcegraph Cody (agent) depends on at runtime — adjacent-infrastructure cell: Sourcegraph search graph; OpenAI + Anthropic + Google APIs
- Speak depends on at runtime — tightly integrated; OpenAI-backed.
- Synthflow depends on at runtime — adjacent-infrastructure cell: Twilio; Deepgram; ElevenLabs; OpenAI / Anthropic
- v0 (Vercel) depends on at runtime — adjacent-infrastructure cell: Vercel deploy; Anthropic + OpenAI + Vercel-trained models
- Vapi depends on at runtime — adjacent-infrastructure cell: Twilio / Vonage SIP; Deepgram + Cartesia + Eleven STT/TTS; Anthropic / OpenAI LLMs
- Vonage AI Studio depends on at runtime — adjacent-infrastructure cell: Vonage CPaaS; Ericsson telco infra; OpenAI / Anthropic / Google Cloud
- Windsurf (Codeium / OpenAI) depends on at runtime — adjacent-infrastructure cell: Anthropic Claude + OpenAI + Codeium-own models
- Zed (Agentic Editing) depends on at runtime — adjacent-infrastructure cell: Anthropic Claude + OpenAI + local models (Ollama)