In [24]: # anthropic_salary.ipynb
Direct answer: Anthropic ML engineer salary
Anthropic's own US job postings list the annual salary range for each role. On 17 October 2026, 14 of its 32 research engineer, research scientist and machine learning postings listed $500,000 - $850,000, the most common range, and 29 of 77 software engineer postings listed $405,000 - $485,000. Equity comes on top; Anthropic does not publish the amount.
$500,000 - $850,000
Research / ML, most common
$405,000 - $485,000
Software eng., most common
not published
equity amount
Abstract
Anthropic is one of the largest frontier AI labs, and it publishes a salary range on its US job postings. Because Anthropic states those ranges itself, they are the one part of an Anthropic package that can be checked. The other part, equity, is set per offer and not published, so this page reports the posted ranges [1] and explains what to ask about the rest, including how the planned IPO changes the value of that equity.
1 Ranges read from Anthropic's public job board on 17 October 2026. Counts cover every US-located posting (San Francisco, New York, Seattle, Washington DC, remote US) with a published range; "research" means a title containing Research Engineer, Research Scientist, Machine Learning or ML (people-research and life-sciences roles excluded). Each posting states an "Annual Salary" range for the role. Postings open and close daily, so the counts are a dated snapshot, not a permanent band. This is not compensation advice.
table an-1 : 17 October 2026
Table an-1. Selected Anthropic US postings and the annual salary range each one states, read 17 October 2026; each role links to the posting. Wide ranges let one posting hire across several seniority levels. Anthropic does not publish an L4 to L7 ladder, so no level mapping is shown. For OpenAI's posted ranges, see the OpenAI vs Anthropic salary side-by-side.
section an-2 : the unpublished half
Anthropic's postings mention competitive compensation and benefits and optional equity donation matching, but give no grant size, instrument or vesting terms. Those are set in each offer letter, which means the equity in an Anthropic package is the part no public source can benchmark. Self-reported packages on crowd-sourced sites are unverified, so they are not restated here.
What you can do is ask the right questions. On any Anthropic offer, get in writing: the number of shares or units granted, the instrument (restricted stock units or stock options, and if options, the strike price and whether they are ISOs or NSOs), the vesting schedule and cliff, the valuation the grant is priced against, and the recent history of employee tender offers. Options are taxed on the spread at exercise and RSUs as ordinary income when they vest (or, for double-trigger RSUs, when the liquidity condition is met), so the instrument changes when the tax bill arrives.
section an-3 : liquidity outlook
Anthropic's reference valuation has risen steeply, which is why the paper value of equity granted at different times differs so much. Its Series G round in February 2026 priced the company at $380 billion post-money ($350 billion pre-money on a $30 billion raise), and its Series H round in May 2026 at $965 billion post-money. Between rounds, Anthropic has run employee tender offers (Bloomberg reported one at a $350 billion valuation in February 2026); participation is at the company's and investors' discretion.
Anthropic filed confidentially for a US IPO on 1 June 2026, with Morgan Stanley, Goldman Sachs and JPMorgan reported as lead underwriters. September 2026 reporting pointed first to an October and then to a November Nasdaq listing. As of 17 October 2026 no shares have priced and Anthropic has not confirmed a date, so any public-market valuation is speculation until the listing completes.
For an employee holding Anthropic equity, a completed IPO would create the first broad liquidity event, replacing discretionary tender offers as the main way to sell vested equity, and would typically carry a lock-up period after listing before employees can sell.
section an-4 : common questions
How much does an ML engineer at Anthropic make in 2026?
Anthropic publishes an annual salary range on its US job postings. On 17 October 2026, 14 of its 32 US research engineer, research scientist and machine learning postings listed $500,000 - $850,000, the most common range; the full set ran from $300,000 at the bottom of the lowest range to $850,000 at the top. These are salary ranges for the posted role, not total compensation: Anthropic offers equity on top but does not publish the amount on its postings.
What is the software engineer salary at Anthropic?
On 17 October 2026, 29 of Anthropic's 77 US software engineer postings listed an annual salary range of $405,000 - $485,000, the most common range; staff-level postings such as Staff Software Engineer, Code RL list up to $625,000.
What is Anthropic total compensation including equity?
Anthropic does not publish total compensation or equity amounts. Its postings give the salary range only, and equity grant size, instrument, vesting schedule and liquidity terms are set per offer. Crowd-sourced sites collect self-reported packages, but those are unverified submissions, so this page does not restate them as a market rate. On any offer, ask for the number of shares or units, the instrument (RSU or option), the vesting schedule and cliff, the most recent valuation the grant is priced against, and the history of employee tender offers.
Does Anthropic pay by level (L4, L5, L6, L7)?
Anthropic does not publish a levelling ladder or level-specific pay bands, and its postings do not carry L-numbers. Its postings use titles such as Software Engineer, Staff Software Engineer, Staff+ Software Engineer and Research Engineer, and many cover a wide range (for example $500,000 to $850,000 for Research Engineer, Machine Learning (Reinforcement Learning)) so that one posting can hire across several seniority levels. Any L4 to L7 table for Anthropic is a third-party mapping, not Anthropic's own structure.
How does Anthropic's planned IPO affect equity compensation?
Anthropic filed confidentially for a US IPO on 1 June 2026, with Morgan Stanley, Goldman Sachs and JPMorgan reported as lead underwriters. September 2026 reports pointed first to an October and then to a November Nasdaq listing; as of 17 October 2026 no shares have priced and Anthropic has not confirmed a date. Its most recent private round (Series H, May 2026) valued the company at $965 billion post-money, up from $380 billion at the Series G in February 2026. For an employee, a completed IPO would create the first broad liquidity event, replacing discretionary tender offers as the main way to sell vested equity, and would typically carry a lock-up period after listing.
Does Anthropic pay more than OpenAI for ML engineers?
On posted salary, yes for most roles we checked. On 17 October 2026, Anthropic's most common US research engineer range was $500,000 - $850,000 against OpenAI's $380,000 - $500,000, and its most common software engineer range was $405,000 - $485,000 against OpenAI's $266,000 - $445,000. Both labs offer equity on top and neither publishes the amount, so posted salary is the only like-for-like comparison available. See the side-by-side page for the detail.
Do you need a PhD to work as an ML engineer at Anthropic?
No. Anthropic's posting for Research Engineer, Machine Learning (Reinforcement Learning), checked in October 2026, lists formal certifications or education credentials, and academic research experience or publication history, under what strong candidates need not have. It states a minimum education of a bachelor's degree or an equivalent combination of education, training and experience. A PhD is more common on research scientist roles.
OpenAI vs Anthropic
Side-by-side posted salary ranges and equity
OpenAI salary deep dive
Single-lab: posted salary ranges, PBC equity, IPO
Frontier-lab tier overview
T1 deep dive, equity forms
Total comp breakdown
How to value an equity package
Negotiation playbook
Frontier-lab offer tactics
RLHF engineer salary
Post-training at frontier labs
Bay Area metro
Where Anthropic HQs
In [ ]: # reader_questions.ipynb
Write in. A person opens every message, and the ones we can answer come back with the survey release or the public disclosure the number is drawn from, stated plainly enough that you can go and check it. We put no clock on a reply. Should an answer be worth reading twice we might set it out here stripped of anything identifying, and we will write and ask you before that happens.
Out[ ]:
Treat everything on this site as compensation reference assembled from published surveys and public filings, not as financial advice and not as a benchmark any single negotiation should be anchored to. Your own offer turns on the employer, the level it is pitched at, the equity instrument behind it and the market on the day it is made.