AI engineer salary in 2026: what companies pay in the US and Europe

In the US, AI companies post a median base band of about $211,000 to $280,000 for AI engineers. The midpoint of a typical band is $256,500, from 39 postings at 12 companies on Tokn as of 27 October 2026. Those are base salaries: equity and bonuses come on top. Salary sites quote much lower averages, because they measure something else.

Below: what employers post, company by company, with sample sizes. The US comes first, where pay ranges are published by law, then Canada and Europe. Looking rather than benchmarking? Browse AI engineer jobs.

In short

• US AI engineer bands on Tokn: median $211k–$280k base, midpoint $256,500 (39 postings, 12 companies). Weighted by company instead of posting: $241,500.
• Posted bands are base pay. GitLab says its range excludes "bonuses, equity, or benefits"; OpenAI lists "Offers Equity" next to its ranges.
• Staff and principal titles post about $40k more at the midpoint than mid-level and senior ones.
• At the same companies, ML engineer and research roles post higher bands than AI engineers.
• Europe: few published bands. Anthropic posted £225k–£255k in London; most other employers posted £80k–£150k or €55k–€140k.

What counts as an AI engineer

An AI engineer builds products on top of existing models (LLM apps, agents, retrieval, evaluations), while a machine learning engineer trains the models and a researcher invents new ones. For every Tokn figure below, the job title decides: a posting counts when its title says AI engineer, Applied AI engineer, AI software engineer, AI product engineer, GenAI engineer, LLM engineer, agentic AI engineer or agents engineer (Scale AI's "Frontier Agents Engineer").

Forward deployed titles count when they also say AI engineer ("AI Engineer – Forward Deployed Engineering"). Plain "Forward Deployed Engineer" titles are a separate comparison group, covered in our forward deployed engineer guide. Managers, directors, interns and titles that say research, infrastructure or support are excluded, and team names don't count: a "Backend Engineer" on an AI team is a backend engineer.

No live posting on Tokn is titled "LLM engineer": employers say "Applied AI engineer" for the same work, described in what an LLM engineer does. Agent-focused roles are under agentic AI jobs.

AI engineer salary in the US: the posted bands

The typical US AI engineer band on Tokn runs from $211,000 to $280,000 in base salary. Half of the band midpoints fall between $236,250 and $299,000.

One warning before you use the headline. OpenAI and Scale AI account for 18 of the 44 North American AI engineer postings (11 and 7). Weighting each company equally gives $241,500 instead of $256,500. We think the honest answer is "somewhere between the two": the higher figure reflects where AI engineers are hired most, the lower one what a typical company pays.

Applied AI engineer salary by company

The employer moves the band more than the title does. Here is what each company posted, sorted by midpoint median.

Elastic also posted one US AI engineer role. We leave its band out: the posting gave two location-based ranges and our data holds only the lower one, so it would understate the pay.

AI labs and AI-native companies post the highest bands; software companies adding AI features post closer to their usual engineering scale.

Senior and staff AI engineer salary

Staff, principal and lead titles post a midpoint median of $283,500 (11 postings, 5 companies), against $242,250 for mid-level (16) and $243,000 for senior (8). Mid and senior look the same because many AI-lab titles carry no level at all ("Applied AI Engineer, Startups") and get counted as mid, while GitLab sits in the senior group. Eight senior postings is not a statistic.

The cleanest picture comes from one employer. Scale AI posted the same Applied AI role at four levels:

Each step lifts the low end by $23k to $49k, and the bands overlap. Which level you are slotted at decides which band you negotiate inside.

AI engineer salary in San Francisco, New York and remote

San Francisco is the only city with enough postings for a median: 25 postings at 7 companies, midpoint median $260,000, with bands from $180,000 to $405,000. Snowflake's four postings in Menlo Park ($156k–$379.5k) are not included in that count.

Everywhere else, the numbers are too small for a median, so here is what was posted:

• New York: two postings with their own band, Celonis at $162k–$190k and OpenAI (API Platform, Digital Natives) at $251k–$278k. Anthropic's Enterprise band ($200k–$320k) covered San Francisco and New York together.
• Remote US: four postings from three companies, with bands from $108.4k to $300k. They include GitLab ($108.4k–$129.6k) and Databricks's AI FDE role ($182k–$250.2k).
• Washington DC: three public-sector postings, OpenAI at $211k–$302k and Anthropic at $280k–$320k.
• Seattle: no AI engineer posting on Tokn as of 27 October 2026.

You can browse the current openings in AI engineering jobs in San Francisco.

How wide the bands are

The median AI engineer band is $67,000 wide, the top about 1.32 times the bottom. The widest, Perplexity at $220k–$405k, likely spans more than one level: the interviews decide where you land.

Base pay vs total compensation

Posted bands are base salary only. California requires employers with 15 or more employees to include a pay scale in job postings, defined as the salary or hourly wage range they reasonably expect to pay on hire (California Labor Code §432.3). Equity, bonuses and benefits sit outside that range, which is why companies flag them separately.

Total compensation at AI companies can sit well above the band. Levels.fyi reported in its Q3 2025 AI engineer compensation report a median that peaked at $295,000 in March 2024 and was around $260k–$269k in 2025. Treat it as 2025 data, and as total comp, not base.

Why do other sites show much lower numbers? Glassdoor's AI engineer page showed an average base of around $140k, as quoted by other guides in 2026. It averages self-reported salaries from every employer using the title, far beyond AI labs. Tokn tracks AI-heavy companies' posted bands: a different question.

One more trap: the US Bureau of Labor Statistics has no "AI engineer" occupation. Pages that quote a BLS figure for AI engineers are using computer and information research scientists, a different job.

AI engineer vs ML engineer vs research vs FDE pay

AI engineers do not post the highest bands in AI. On Tokn, ML engineer and research roles post more, both across the board and at the same companies.

Method: same title rules, same date, US-dollar bands only, midpoint = (low + high) / 2, managers, directors and interns excluded. Research is concentrated (Anthropic and OpenAI post 90 of the 132 bands), and OpenAI posts 14 of the 30 FDE bands.

The comparison holds at the same employers. At the 9 companies that post both AI engineer and ML or research bands, AI engineers sit at $258,250 (36 postings), ML engineers at $308,700 (15 postings, 5 companies) and research at $397,500. Our reading: roles that work on the model itself post more than roles that build products on it. For the model-training side in Europe, see our guide to machine learning engineer salaries in London, Paris and Berlin.

AI engineer salary in Canada

Tokn has too few Canadian bands to give a figure. Only three AI engineer postings were in Canada as of 27 October 2026, and one had a band: Snowflake's AI Software Engineer in Toronto, at CA$176,300–CA$226,300.

AI engineer salary in Europe

European employers publish far fewer bands, so there is no reliable European median. Tokn could read a salary on 14 of 56 European AI engineer postings, none in US dollars. Here they are, one line per posting and in their own currency.

The fourteenth, Anthropic's Applied AI Engineer, Enterprise role in Paris, showed €220,000–€235,000. That figure is on-target earnings, including commission, rather than base, so it is not comparable with the rows above.

In London, the table shows two levels. Anthropic posted near-US bases (£225k–£255k across three roles); the other employers, senior roles included, posted £80k–£150k. In France and Spain, the published bands run from €55k to €140k, with no AI lab base band to compare against.

Elastic's one role posted in three countries shows location at work: about €135k at the midpoint in Ireland, about €95k in Greece.

The biggest European hirers publish nothing. OpenAI (11 European AI engineer postings) and Mistral AI (7) gave no salary in Europe, though OpenAI publishes bands for the same titles in San Francisco. The EU Pay Transparency Directive was due to be transposed by 7 June 2026, but most EU countries missed the deadline (Ius Laboris tracker). When it applies, it requires employers to give the starting pay or its range before the interview, not necessarily in the ad.

For every role across the continent, see our pillar on AI salaries in Europe, role by role, or browse LLM jobs in London.

How to use these numbers in a negotiation

Use posted bands to anchor your base, and negotiate equity separately. A few rules:

• Ask for the band when it isn't posted. In Europe it often isn't, and the question is normal.
• Find out your level before discussing numbers. On Scale AI's ladder, moving from senior to staff lifts the whole band by $49k to $61k.
• Separate base from equity. Compare offers on base first, then on equity value, vesting and liquidity.
• Expect remote bands to be location-adjusted. The remote US bands on Tokn sit lower than San Francisco ones.
• Compare like with like: a posted base band is not a self-reported total comp figure.

Our guide on how to negotiate an AI salary covers how to read a band and what to say.

Find AI engineer jobs with a salary

About 9 in 10 North American postings on Tokn show a pay range, so you can filter before you apply. Browse all AI engineer jobs, and get a job alert to hear about new roles as companies post them.

Method. Tokn snapshot of active jobs, 27 October 2026. A posting counts as an AI engineer when its title says so (rules above). Figures are posted base pay; midpoint = (low + high) / 2. Medians only where there are enough postings. USD, CAD, GBP and EUR are never converted or mixed.

FAQ

How much does an AI engineer make in the US?

AI companies post a median base band of $211,000–$280,000 for AI engineers, with a midpoint of $256,500 (39 postings at 12 companies on Tokn, as of 27 October 2026). Weighted by company, it is $241,500. Equity and bonuses come on top.

How much do AI engineers make in San Francisco?

The midpoint median of 25 San Francisco AI engineer postings on Tokn is $260,000 in base pay, with bands from $180,000 to $405,000. San Francisco is the only US city with enough postings for a median.

What do OpenAI and Anthropic pay applied AI engineers?

OpenAI posted most of its applied AI engineer bands between $197k–$280k and $251k–$278k, with $211k–$302k for government roles. Anthropic posted $200k–$320k for its Enterprise role and $280k–$320k for its other applied AI roles. Both are base pay; OpenAI lists equity separately.

Do AI engineers earn more than ML engineers?

Not on posted bands. ML engineers post a midpoint median of $276,244 against $256,500 for AI engineers, and at the same nine companies the gap widens ($308,700 vs $258,250). Research roles post the most.

Is the posted range base salary or total compensation?

Base salary. California defines a posted pay scale as the salary or wage range expected on hire, and companies such as GitLab state that the range excludes bonuses, equity and benefits. Your total compensation at an AI company can sit well above the band.

How much do AI engineers earn in London, Paris or Germany?

Few employers publish bands in Europe. Anthropic posted £225k–£255k for London applied AI roles; other London employers posted £80k–£150k, and French and Spanish bands ran from €55k to €140k. Tokn has no published AI engineer band in Germany.