Hugging Face scores 85 out of 100 on ENTRA's August 2026 Pay Transparency Index — the highest mark among 20 scored AI employers, and 3 points ahead of the next entry, Stripe. Only two companies reach the AA tier. (The ENTRA Pay Transparency Index is a distinct five-dimension methodology from the ENTRA 100 overall employer ranking, which scores Hugging Face at 86 AA+; the two indices measure different dimensions.) The remaining 18, a list that includes OpenAI, Anthropic, Google DeepMind, and NVIDIA, score at BBB or below. Twelve score B or BB, meaning proactive disclosure on any measured dimension is largely absent.
The gap between Hugging Face and its peer set is not explained by company size or regulatory pressure. At roughly 700 employees across 50-plus countries, Hugging Face is a fraction of the headcount of the companies it outscores. It operates primarily under French labor law and US state regulations — the same legal scaffolding that every competitor with European and American employees navigates. It has made no public statements claiming a compliance mandate forced its hand. The transparency is a design choice, made explicitly and extended globally, and it shows up in candidate data in ways that matter for anyone deciding where to apply.
The Transparency Score Architecture
The ENTRA Pay Transparency Index scores five weighted dimensions. Hugging Face's 85 breaks down across all five (ENTRA Pay Transparency Audit Q2 2026, data window January–June 2026; posting audit July 2026):
Salary Band Disclosure — 27 of 30 points. The index's highest-weighted dimension rewards companies that post salary ranges globally, not only where state or national law compels them to. Hugging Face publishes ranges on the overwhelming majority of its job postings — including roles based in France, Germany, Singapore, and the United Kingdom — without being legally required to do so in most of those jurisdictions. A candidate in Berlin considering a Hugging Face role sees the compensation band before submitting an application. Competitors in the same city hiring for equivalent research engineering roles typically do not show a number until the recruiter call.
Equity Compensation Transparency — 21 of 25 points. The second-largest dimension measures whether equity terms are legible to candidates before the offer stage. Hugging Face's engineering blog addresses this directly: vesting schedules, cliff mechanics, how grant values are pegged to funding round pricing, and what dilution looks like across hypothetical exit scenarios. Most pre-IPO AI companies treat this as NDA territory until offer acceptance. Hugging Face publishes it as candidate-facing content. The 4-point gap from a perfect 25 reflects the absence of per-level grant range data, which would require a level of specificity the company has not yet made public.
Internal Pay Equity Data Published — 15 of 20 points. Hugging Face's pay equity disclosure reflects internal equity analysis and careers-page language that goes beyond the UK mandatory gender pay gap filing but falls short of an intersectional audit with published methodology. That gap — the distance between affirming pay equity exists and publishing the statistical test that verifies it — is the single largest structural deficit between the company's current 85 and a theoretical AAA score. No company in the August 2026 index reaches AAA. Intersectional audit publication with statistical methodology is what the threshold requires.
Compensation Philosophy Public — 13 of 15 points. Careers documentation at Hugging Face includes explicit reference to compensation philosophy: what market percentile the company targets, how cash and equity balance at different levels, and how the company benchmarks against both European and North American technology markets. Clément Delangue, the co-founder and CEO, has discussed pay philosophy in public — on LinkedIn and in podcast conversations — in a way that is unusual for private-company executives, where compensation philosophy is typically internal. The 2-point gap reflects the absence of a formal, published, versioned document of the kind that earns full marks.
Market-Rate Commitment and External Benchmarking — 9 of 10 points. Hugging Face's careers materials carry specific language about competitive benchmarking against technology market rates in both European and North American geographies. That specificity, rare in AI employer communications, earns the near-maximum score. The single missing point reflects the absence of a named percentile target — a statement of the form "we pay at the 75th percentile of the US technology market" — which would close the gap.
What the Bands Say
The salary data tells the transparency story more concisely than any policy document. A Senior ML Engineer role posted by Hugging Face in the July 2026 audit window carried the following ranges by geography:
- New York: $140,000–$185,000 base salary
- Paris: €128,000–€170,000 base salary (USD equivalent at canonical rate 1.09: $140K–$185K)
- London: £110,000–£145,000 base salary (USD equivalent at canonical rate 1.27: $140K–$184K)
Three jurisdictions, three local currencies, one underlying methodology. The bands translate to near-identical USD values across geographies — a signal of deliberate, consistent approach rather than localized compliance patching. The band width in each case is $45,000 or its equivalent. That figure matters. Anthropic's posted base band for comparable seniority runs $120,000–$320,000 — a $200,000 spread. OpenAI's equivalent runs $100,000 to more than $300,000. A candidate reading an Anthropic or OpenAI posting knows they are somewhere in a range that spans the difference between a mid-career salary and a senior executive base. A candidate reading a Hugging Face posting knows, within $45,000, what they will be offered before the first recruiter call.
Narrow bands carry a practical negotiation implication for candidates: most of the information asymmetry that enables lowball opens disappears. When a recruiter's first offer is constrained to a $45,000 range that the candidate has already seen, the conversation starts 45 percent closer to resolution than it does when the candidate is calibrating against a $200,000 range they saw before ever speaking to a human.
The equity picture is more complex, because Hugging Face remains private at a $4.5 billion valuation. The Series D, led by GV (Google Ventures) with participation from Alphabet, closed at that mark in 2023. Candidates evaluating equity grants must hold two facts in tension: the valuation has appreciated substantially from earlier rounds, which compresses near-term paper gains for new grantees, and the open-source infrastructure thesis — Transformers library, the Hub, 800 million-plus monthly model downloads across half a million public models — gives the company a market-position argument that is difficult for competitors to replicate. The engineering blog content on equity mechanics, which earns most of Hugging Face's equity transparency score, is the tool candidates need to model this trade-off themselves before entering an offer conversation.
The GGML.ai acquisition, completed in February 2026, added inference optimization talent — the team behind the GGML quantization format and llama.cpp ecosystem infrastructure. Compensation parity was maintained for acquired team members; ENTRA's Q2 recruiter network data, drawn from placement conversations across the inference engineering category, found no evidence of post-acquisition band compression or renegotiation.
The Talent Effect
Transparency at the band level is a design choice. What it produces at the candidate pipeline level is measurable.
ENTRA's Q2 2026 Candidate Survey (n=340, Hugging Face candidates who received or declined offers H1 2026) found that 73 percent cited salary transparency as a top-three reason for applying to Hugging Face specifically, ahead of a role at a non-disclosing peer. The survey controls for role type and seniority band; the transparency effect holds across research engineering, applied AI, and developer relations functions. For candidates at the senior-IC level — the tier where Anthropic, OpenAI, and xAI compete most aggressively — the transparency preference is even more pronounced: 81 percent of senior candidates in the sample identified visible bands as a top-three driver of first-round application completion.
The retention signal is the more durable number. ENTRA's Talent Index tracks annualized attrition across 20 scored AI employers using a composite of LinkedIn tenure data, Glassdoor review signals, and recruiter network intelligence. Hugging Face's attrition rate runs 18 percent below the non-transparent peer set — companies in the BB and B tiers of the Pay Transparency Index — after controlling for company size and tenure distribution. The causal mechanism is not definitively established, but the directional interpretation is intuitive: employees who accepted an offer understanding its full structure are less likely to feel misled when their equity model or raise conversation plays out as described.
Thomas Wolf, the company's Chief Science Officer and the architect of much of the Transformers ecosystem, has been the public face of Hugging Face's open-source community posture. That same instinct toward openness — document what you're doing, make the knowledge available — extends to how the company talks about money. The compensation philosophy is not a separate culture decision from the technical one. For Clément Delangue and Julien Chaumond, the founding frame is: transparency is load-bearing infrastructure, not marketing.
What It Signals for the Industry
Hugging Face's 85 is a proof of concept for a different kind of employer brand: one built on information rather than aspiration, legible before the first conversation and consistent across New York, Paris, and London simultaneously. Two AA companies in an index of 20 are a constraint, not a ceiling — and the distance between 85 and the first company below the AA line is 3 points, not 30.
The candidates who understand what that 3-point gap costs the rest of the field are the ones who know which number to ask for before the call begins.
ENTRA Pay Transparency Index August 2026 methodology: five weighted dimensions — Salary Band Disclosure (30 pts), Equity Compensation Transparency (25 pts), Internal Pay Equity Data (20 pts), Compensation Philosophy (15 pts), Market-Rate Commitment (10 pts); 20 AI employers scored; data window January–June 2026; posting audit July 2026; AA tier: 80–89; AAA tier: 90–100. ENTRA Q2 2026 Candidate Survey: n=340 Hugging Face candidates who received or declined offers H1 2026; controlled for role type and seniority band. ENTRA Talent Index attrition methodology: composite of LinkedIn tenure data, Glassdoor review signals, and recruiter network intelligence; annualized attrition compared across 20 scored AI employers. Salary band data: Hugging Face Senior ML Engineer bands (New York $140K–$185K, Paris €128K–€170K, London £110K–£145K) per ENTRA Q2 2026 posting audit, July 2026 window; EUR/USD at 1.09, GBP/USD at 1.27 (ENTRA August 2026 canonical rates). Competitor band comparison: Anthropic Senior ML Engineer base $120K–$320K and OpenAI equivalent $100K–$300K+ per ENTRA Q2 2026 posting audit and Levels.fyi self-reported data, cross-referenced. Hugging Face Series D: led by GV (Google Ventures) with participation from Alphabet; $4.5B valuation; closed 2023; source: company announcement and Crunchbase. GGML.ai acquisition: completed February 2026; compensation parity maintained per ENTRA Q2 2026 recruiter network data. Hugging Face employee count (~700), countries (50+), monthly model downloads (800M+), public model count (500K+): per Hugging Face company statements and Hub statistics, Q2 2026. Thomas Wolf role: Co-founder and Chief Science Officer; architect of the Transformers library ecosystem. Individual named companies (Anthropic, OpenAI, Stripe) did not confirm ENTRA Pay Transparency Index scores.
Global Career PlatformFind AI talent. Find your next role.
Booking is hotels. · Airbnb is apartments. · ENTRA is global careers.