Job Market Data

OpenAI Posted 822 Jobs With Pay Up to $850,000

OpenAI listed 822 jobs on November 12, 2026, with research pay up to $850,000 and average equity near $1.5 million per employee. Hiring runs three to eight weeks.

Working at OpenAI in 2027: Culture, Pay and How to Get Hired
Working at OpenAI in 2027: Culture, Pay and How to Get Hired — StartupStockPhotos / pixabay

OpenAI had 822 open roles on its job board on November 12, 2026, with posted US base salaries reaching $850,000 for research positions. That's before equity. The Wall Street Journal put average stock-based compensation at roughly $1.5 million per employee in 2025.

The company employed about 4,500 people in March 2026, per the Financial Times, and targets roughly 8,000 by year-end. That's near-doubling headcount in nine months. It also means fierce competition per seat. The takeaway: the hiring volume is real, but so is the selectivity.

What does OpenAI actually pay?

Posted ranges on November 12, 2026, each spanning several levels, exclude equity:

  • Software engineer: $145,800–$536,000 (most $266,000–$445,000); London £194,000–£280,000
  • Researcher / research scientist: $170,000–$850,000; London £170,000–£445,000
  • Product manager: $266,000–$530,000
  • Data scientist: $265,000–$515,000
  • Security engineer: $266,000–$510,000
  • Sales, policy, legal, marketing, recruiting: roughly $162,000–$445,000 depending on family

Equity is the differentiator. OpenAI reportedly scrapped its vesting cliff (the waiting period before shares start accruing) in late 2025, so new hires vest from day one. But CEO Sam Altman said in September 2026 there will be no IPO this year, so shares stay illiquid except through company-run sales — ask your recruiter how those work.

Since the October 2025 restructure, employees hold equity in OpenAI Group PBC, a public benefit corporation, rather than the old capped-profit vehicle. Benefits include medical, dental and vision cover, a matched 401(k), 20 weeks of paid parental leave, a learning stipend and daily meals.

Where are the jobs?

Go-to-market leads with 180 openings, then applied AI (123), scaling and infrastructure (93), research (54), user operations (50), marketing (41), forward deployed engineering (39) and security (32). San Francisco lists 575 roles, New York 118, Seattle 80, London 44, plus smaller hubs in Singapore, Tokyo, Washington DC and Europe. Most roles are hybrid at three office days a week; some research and leadership roles require four or five.

What does the interview process look like?

OpenAI's official interview guide outlines five stages, and candidates from 2025–2026 report a full timeline of three to eight weeks:

  1. Application — recruiting replies in about a week
  2. Introductory calls — recruiter or hiring manager, 30–45 minutes each
  3. Skills assessment — pair coding, take-home or technical test, sometimes more than one
  4. Final interviews — 4–6 people over 1–2 days, usually virtual
  5. Decision — references and offer, typically within a week

Engineers face system design at real scale: inference serving, distributed training, high-traffic chat backends. Researchers walk through past papers and frame an open problem. Go-to-market candidates get case studies. Behavioral questions cover ownership, ambiguity and AI's impact.

On AI tools in interviews: there's no published rule. Research and infrastructure coding rounds generally run without AI help; some applied discussions allow it. Assume no assistance unless your recruiter says otherwise.

How does it compare with Anthropic?

Anthropic had about 5,300 staff in September 2026 and 674 open roles on November 12, versus OpenAI's 4,500 and 822. Anthropic wants staff in office at least 25% of the time and filed an S-1 in October 2026 (Nasdaq: ANT). OpenAI emphasizes speed and ownership in interviews; Anthropic adds a dedicated values round.

OpenAI also runs a six-month, full-salary Residency for researchers and engineers moving into AI from fields like mathematics, physics or neuroscience. No formal ML background is required. The 2026 cohort's applications have closed.

The company says it isn't credential-driven — strong, demonstrable work beats degrees. Candidates who target one or two closely matching roles, quantify their impact and arrive with a real position on AI's risks and benefits consistently outperform broad applicants. Expect the pace to stay relentless as headcount doubles.

Original: openai.com

openai-hiringai-salariestech-job-openingsopenai-interview-processemployee-equity

Share this article:

More from Tom Whitfield

Tom Whitfield

Show full bio

Senior reporter covering regulation, policy and their impact on Job market data and labor economics at Path & Pay.

591 articles

Go deeper

« Previous articleNext article »