Consolidating data points I've collected from a few different sources over the last 6 months. All for SWE, senior (L5/E5 equivalent) unless noted. SF Bay Area unless noted remote.
Data points:
Offer A (senior SWE, infra, SF, early 2026): $230k base, $1.2M equity over 4 years (25% cliff), $30k signing. No recurring cash bonus. Total year-1 comp including signing: roughly $560k.
Offer B (senior SWE, model training tooling, SF, late 2025): $220k base, $1.1M equity over 4 years. Recruiter said they don't do recurring performance bonuses the way Google does, the compensation philosophy is to front-load base and equity instead.
Offer C (senior SWE, safety, SF, early 2026): $210k base, $1.4M equity. Slightly lower base, higher equity. This person told me they got bumped on equity specifically by referencing another senior AI lab offer.
Remote offer D (senior SWE, product eng, fully remote, 2026): $200k base, $900k equity. Remote discount is real but smaller than some companies.
Patterns:
Base at senior level: $200-235k. Equity at senior: $900k-$1.5M over 4 years, wide variance. The equity range is the main lever in negotiation. No standard annual cash bonus structure, though there are informal discretionary ones.
Refresh grants exist but the cadence and size vary. One person told me they got their first refresh at the 18-month mark, another said annual starting at year 2.
Equity is in the form of restricted stock, but it's pre-IPO so the actual value is based on 409A valuations and secondary market. As of early 2026 those secondary valuations were meaningfully higher than the last primary round, but illiquid as always. Weight accordingly.
Leveling: Anthropic doesn't publish level names publicly but internally they use a numeric ladder. Senior SWE maps to roughly L5 at Google or E5 at Meta in terms of the experience bar they set.
All comp unverified by me directly; treat as directional.