OpenAI
Updated May 2026
GPT / ChatGPT
The household name that kicked off the modern AI era
OpenAI began as a nonprofit research lab in 2015 and became the company everyone associates with AI after ChatGPT launched in November 2022. Their philosophy has shifted from 'publish all research' to 'scale quickly, deploy widely, keep the frontier.' In 2026 the GPT-5 family spans frontier reasoning, coding, computer-use workflows, small high-volume models, and Codex-optimized agentic coding models.
Variants
11
Best at
broadest ecosystem and tooling
Max context
1.1M
tokens
Pricing
Free
$0
per ChatGPT with rate limits
Plus
$20
per month
Pro
$200
per month, unlimited
GPT-5.5 API
$5 in / $30 out
per million tokens
GPT-5.4 API
$2.50 in / $15 out
per million tokens
GPT-5.4 mini API
$0.75 in / $4.50 out
per million tokens
GPT-5.4 nano API
$0.20 in / $1.25 out
per million tokens
GPT-5.3-Codex API
$1.75 in / $14 out
per million tokens
Variants
Sort the table by context window or cost to find the right variant. Click any version below for a battle card with ranks, pricing notes, and official links.
| Modalities | ||||
|---|---|---|---|---|
GPT-5.5 Pro gpt-5.5-pro | 1.1M | varies / varies | 2026 | textvisioncode |
GPT-5.5 gpt-5.5 | 1M | varies / varies | 2026 | textvisioncode |
GPT-5.4 mini gpt-5.4-mini | 400K | varies / varies | 2026 | textvisioncode |
GPT-5.4 nano gpt-5.4-nano | 400K | varies / varies | 2026 | textvision |
GPT-5 gpt-5 | 400K | $10 / $40 | 2025 | textvisioncode |
GPT-5.3-Codex gpt-5.3-codex | 400K | varies / varies | 2026 | textvisioncode |
o3 / o4-mini o3 | 200K | $10 / $40 | 2025 | textvision |
GPT-4o gpt-4o | 128K | $2.50 / $10 | 2024 | textvisionaudio |
GPT Image 2 gpt-image-2 | — | varies / varies | 2026 | image |
Sora 2 sora-2 | — | varies / varies | 2025 | video |
Whisper whisper-1 | — | varies / varies | 2022 | audio |
Battle card
Context rank
#2
within GPT / ChatGPT
Capability rank
#1
modalities + reasoning
Weights
Closed
API/app access
Best fights to pick
- complex software engineering
- professional reasoning
- multi-tool research
- high-stakes planning
Rankings are Tendril directory ranks, computed from the model data shown here. Public benchmark leaderboards change often, so official docs and current benchmark pages should be checked before buying or deploying.
Learn
Lessons about this model
Structured lessons that cover GPT / ChatGPT directly or put it in context alongside its rivals.
OpenAI Model Picker: GPT-5.5, GPT-5.4, Mini, Nano, and Codex
Codex: The Map of OpenAI's Coding Agent
Reviewing Codex Output Like a Senior Engineer
The Responses API: OpenAI's Modern Developer Surface
OpenAI Tool Use: Functions, Web Search, Files, MCP, Shell, and Computer Use
Structured Outputs: Make the Model Return Data You Can Trust
GPT-5.5 vs. GPT-5.4 mini — when to pay for the flagship
GPT-5.5 vs. Claude Opus 4.7 — which chatbot wins your day
Claude Haiku 4.5 vs. GPT-5.4 mini — the cheap-and-fast class
Reasoning effort — when to pay for deeper thinking
ChatGPT Agents — OpenAI's Operator, matured
ChatGPT Memory: The Feature That Made ChatGPT Personal
Custom GPTs: Shareable ChatGPTs Anyone Can Make
OpenAI API basics
ChatGPT, November 2022
GPT-3 and the Scaling Laws
Check yourself
Quizzes
Short, mixed-difficulty quiz sets on GPT / ChatGPT and its model family.
Hands-on
Try these prompts
Ready-made prompts that show GPT / ChatGPT at its best. Use them in your own AI workspace, then compare the output with what you learned in Tendril.
GPT-5 on structured output
BuildersAsk GPT to return strict JSON — it's one of the most reliable models for 'just give me the schema' extraction.
Extract any people, companies, and dollar amounts from this paragraph into JSON with keys people, companies, amounts. Only return the JSON, no prose. 'In January 2025, Sarah Chen at Vanta announced a $40M Series B led by Sequoia, with participation from Y Combinator and a $2M follow-on from angel investor Marcus Green.'
High-effort reasoning on a word problem
CreatorsSend a math problem to a high-effort reasoning model and check the visible answer and math carefully.
A train leaves station A at 2pm going 60mph toward station B, 200 miles away. At 3pm a second train leaves station B going 80mph toward A. At what time do they meet, and how far from A? Show your work.
GPT for rapid persona writing
BuildersGPT handles voice and tone transfer quickly — great for marketing drafts.
Rewrite this feature announcement three times: once in the voice of an enthusiastic founder, once as a dry enterprise PR release, once as a Twitter thread by someone trying to go viral. 'We launched a new dashboard that shows your weekly spending by category.'
Print & keep
Printable reference
One-page summaries and flowcharts — great for desks, classrooms, or study sessions.
Go deeper
Official resources
Straight from the lab — docs, API references, and the chat surfaces you can try today.
Strengths
- broadest ecosystem and tooling
- strong reasoning in GPT-5.5 and GPT-5.4
- multimodal image/video/audio coverage
- Codex coding-agent workflow
Limits
- closed weights
- changing model lineup confuses users
- tool access requires careful permissions