OpenAI

OpenAI

Overview

[[Sam Altman]] and a coalition of AI researchers and technologists founded OpenAI in 2015 as a nonprofit organization dedicated to ensuring artificial general intelligence benefits humanity. The organization started with ambitious goals around safety and open collaboration, but evolved significantly over the past decade. In 2023, OpenAI transitioned to a capped-profit structure while maintaining its nonprofit parent, enabling faster capital raises and scaling while nominally preserving safety governance. Today, OpenAI stands as the market leader in frontier AI capabilities and consumer AI adoption, having captured the public imagination with models like GPT-4 and now GPT-5.5.

As of May 2026, OpenAI has reportedly closed a $122B funding round at an $852B post-money valuation — described in coverage as the largest private financing in history. The round is anchored by Amazon, NVIDIA, SoftBank, and Microsoft. The shipment cadence has accelerated through 2026: GPT-5.4 launched in March, GPT-5.5 followed in April with a 1-million-token context window and agentic desktop workflows, and GPT-5.5 Instant became the new default ChatGPT model in early May. OpenAI also discontinued the standalone Sora app in late March 2026, conceding ground in consumer video generation to Google's Veo line.

Key Details

  • Founded: 2015
  • Founders: [[Sam Altman]], [[Elon Musk]] (departed 2018), [[Ilya Sutskever]] (departed 2024), and others
  • CEO: [[Sam Altman]]
  • Headquarters: San Francisco, CA
  • Funding / Valuation: ~$122B funding round closed (May 2026) at $852B post-money valuation
  • 2025 Revenue: $13.1B
  • Website: https://www.openai.com

Current Models

  • [[GPT-5.5]] — Current flagship; 1M-token context window, agentic desktop workflows, 75% on OSWorld-V (above the 72.4% human baseline); $5/$30 per million input/output tokens (April 2026)
  • GPT-5.5 Pro — Highest-capability variant of GPT-5.5; $30/$180 per million input/output tokens
  • [[GPT-5.5 Instant]] — New default ChatGPT model (May 5, 2026); 52.5% fewer hallucinated claims vs. GPT-5.3 Instant on high-stakes prompts; ~30% more concise; AIME 2025 score 81.2; personalization across past chats, files, and Gmail
  • GPT-5.4 — Pro and Thinking variants; previous flagship; 256K context window (March 2026)
  • [[GPT-5.2]] — Earlier 2026 flagship; 400K context; Instant/Thinking/Pro modes
  • [[GPT-5.2 Codex]] — Specialized code variant
  • [[GPT-5.5-Cyber]] — Cyber-permissive variant of GPT-5.5 for vetted defenders; limited preview launched May 7, 2026 under the "Trusted Access for Cyber" framework
  • [[GPT-4.1]] — Previous-generation flagship model
  • [[GPT-4.1 mini]] — Efficient smaller variant of GPT-4.1
  • [[o4-mini]] — Current small reasoning model
  • [[o3-mini]] — Previous-generation small reasoning model

Key People

  • [[Sam Altman]] — CEO, steers company strategy and investor relations
  • [[Ilya Sutskever]] — Co-founder and former Chief Scientist (departed 2024 to found Safe Superintelligence Inc.)
  • [[Elon Musk]] — Co-founder (departed board 2018, now competing via [[xAI]])
  • [[Reid Hoffman]] — Early backer and board-linked investor

Recent Developments

  • GPT-5.5-Cyber Limited Preview (May 7, 2026): OpenAI launched GPT-5.5-Cyber as the most permissive model in its cybersecurity lineup — designed for defenders to write proofs of concept, hunt for bugs, study malware, reverse engineer attacks, and run controlled red-team simulations. Hard-blocks remain on credential theft and malware authoring. Available only to vetted partners under the "Trusted Access for Cyber" framework; Advanced Account Security required from June 1, 2026. The release is OpenAI's direct response to Anthropic's [[Claude Mythos Preview]] (April 7) and lands inside an intensifying federal-oversight conversation around frontier-class cyber capability.
  • Trump Admin AI Pre-Release Testing (May 5, 2026): Microsoft, Google, and xAI agreed to let the U.S. government test their frontier AI models before launch. OpenAI was already inside the federal pre-deployment evaluation framework via the U.S. AISI partnership.
  • GPT-5.5 Instant Default (May 5, 2026): GPT-5.5 Instant became the new default model for ChatGPT, replacing GPT-5.3 Instant. OpenAI reported 52.5% fewer hallucinated claims than GPT-5.3 Instant on high-stakes prompts (medicine, law, finance) and 37.3% reduction in inaccurate claims on flagged conversations. Responses run ~30% shorter (fewer words and lines) with fewer gratuitous emoji. The model can now reach back into past conversations, files, and Gmail (for Plus/Pro users on web initially) to give more personalized answers, and OpenAI introduced "memory sources" to show which context drove personalization. GPT-5.3 Instant remains available to paid users for three months before retirement.
  • $122B Funding Round Closed (May 2026): Reported close at $852B post-money valuation — Amazon, NVIDIA, SoftBank, Microsoft anchored. Largest private financing on record.
  • Pentagon Classified Deployment (May 1, 2026): OpenAI cleared as one of the seven (now eight) Big Tech vendors to deploy AI on DoD Impact Level 6/7 classified networks via the GenAI.mil portal.
  • GPT-5.5 Launch (April 24, 2026): 1M-token context window, autonomous multi-step desktop workflows, 75% on OSWorld-V agentic benchmark (vs. 72.4% human baseline). Pricing: $5/$30 per million input/output tokens for GPT-5.5 and $30/$180 for GPT-5.5 Pro. Long-context retrieval scored 74% (vs. 36.6% on GPT-5.4) — more than double.
  • GPT-5.4 Launch (March 5, 2026): Pro and Thinking variants released with 256K context window, an iteration over the GPT-5.2 family and a stepping-stone to GPT-5.5.
  • Sora App Discontinued (Late March 2026): OpenAI discontinued the standalone Sora consumer app, ceding the production-video-generation space to Google Veo 3.1 / Veo 4.
  • Revised Compute Spending Plans: Scaled back projected compute infrastructure spending through 2030 from $1.4 trillion to $600 billion, reflecting revised growth expectations and efficiency improvements.
  • Projected Financial Trajectory: 2025 revenue of $13.1B with expected 2026 losses of $14B, but projecting $280B revenue by 2030.
  • IPO Preparation: Preparing for potential initial public offering in Q4 2026.
  • Leadership Transition Impact: Departure of [[Ilya Sutskever]] to found Safe Superintelligence Inc. in 2024 marked a significant leadership shift.

Why They Matter

OpenAI remains the gravitational center of the AI industry, wielding outsized influence over AI development trajectory and public perception. GPT-5.5's 1M-context window and OSWorld-V results — crossing the human baseline on a credible agentic-workflow benchmark — are widely read as the clearest sign yet that frontier models are transitioning from "chat tools" to autonomous digital coworkers. The $122B raise at $852B post-money valuation underscores enduring investor conviction despite mounting losses. However, OpenAI faces a critical inflection point: profitability remains elusive, the standalone Sora app was retired in late March, and the lab is increasingly under pressure from Anthropic on safety positioning, Google DeepMind on multimodal/video, and Reflection AI on open-weight frontier capability. The departure of [[Ilya Sutskever]]—one of the lab's safety-oriented co-founders—to launch a competing safety-focused company signals lingering internal tensions between capability scaling and responsible development. OpenAI's potential 2026 IPO will determine whether the nonprofit structure survives that transition.

Last Updated

May 10, 2026

→ Back to Companies