Orbyt vs Levels.fyi.

Levels.fyi built the community. Orbyt built the pipe.

Community-submitted tech compensation database

At a glance.

Infrastructure coverage.

Orbyt

Levels.fyi

What Levels.fyi is.

Levels.fyi is a community-submitted tech compensation database that anyone can browse in a web browser. Founded in 2017 by Zuhayeer Musa and Zaheer Mohiuddin, it grew into the default reference for candidates weighing FAANG offers. The product is a consumer lookup tool: type a company, pick a level, read a total-comp range pulled from verified employee submissions. That is the full feature set for the public product. There is no documented public API. There is no MCP manifest. There is no CC BY license on the underlying data. Levels.fyi is excellent at what it is: a browser tab you open, read, and close. Orbyt Intelligence is a different category of product: a public salary data API with programmatic access, agent support, and a redistribution license.

Pricing, head to head.

	Entry	Higher tiers
Levels.fyi	Lookup only	15% of raise (success-based)
Orbyt Intelligence	60 req/min free, 300 on Pro ($99/mo)	$199/mo (Ultra)

Levels.fyi's public lookup is free and the comp data is browsable without an account. The revenue model is a negotiation coaching service that charges 15% of any raise they help you secure, plus a private enterprise benchmarking tier that is quote-only. There is no developer tier, no published API pricing, and no self-serve path to programmatic access. Orbyt Intelligence takes the opposite posture: three published tiers starting at $0. Free unlocks 60 req/min and 1,000 requests a month with Bearer authentication and the AI Role Taxonomy engine, no card. Pro at $99/mo unlocks 300 req/min, the MCP server with six locked agent tools, and the /lineage provenance endpoint. Ultra at $199/mo unlocks 1,500 req/min plus the company leveling catalog and the annual compensation reports. For a candidate reading a single number, Levels is free and fast. For anyone building a product, Levels has no pricing because Levels has no product for you to buy.

Orbyt wins.

A lookup, or a pipe.

Structured total compensation: base, equity, bonus and signing as separate fields, plus the published BLS wage for the occupation the role maps to. Levels shows one number; Orbyt shows the structure underneath and what a government source printed.

MCP server with six locked tools. Decision-Ready Responses with quotable answer + citation + follow-up suggestions. Drop into Claude Code, ChatGPT, any MCP client.

A lineage API, not a black box. Every estimate carries a data_point_id, and UK and Canada figures resolve through /lineage to the full source trail with weights and reconciliation rules. US figures are served from a modeled baseline today and the methodology page says so. Levels has no source attribution surface at all.

Public API with Bearer authentication. 60 req/min free, 300 req/min on Pro at $99/mo. Self-serve. No procurement.

Forward projections through 2030 with methodology disclosure

CC BY 4.0 public data license. Cite, redistribute, build on top of.

Orbyt Intelligence is compensation data as infrastructure. A public API. An MCP manifest for AI agents. Forward projections through 2030. A CC BY 4.0 license you can cite, redistribute, and build on. Everything a salary dataset becomes when it is designed to be built on, not just read.

Feature by feature.

Feature by feature, the products overlap on exactly one axis: both show tech salary data in a web UI. Everything else diverges. Levels.fyi covers roughly 200 roles, weighted heavily toward software engineering at the top 50 tech companies, with the deepest data in major metros. Orbyt Intelligence covers 3,445 roles across 81 U.S. cities, spanning tech, finance, design, product, operations, marketing, sales, legal, and public sector. Levels updates when submissions come in. Orbyt updates on a fixed quarterly pipeline with dated source citations (BLS OES, and H-1B LCA from the DOL). Levels shows today's number. Orbyt projects forward through 2030 with published methodology. Levels has no API. Orbyt publishes an OpenAPI 3.1 spec and an MCP manifest. Levels data is proprietary. Orbyt data is CC BY 4.0. On the features a developer or analyst actually needs: API, MCP, projections, license, breadth, uniform coverage: Levels ships none and Orbyt ships all six.

Who each is for.

Use Levels.fyi if you are a senior software engineer weighing a FAANG offer and you want the tightest brand-level comp data available. For L5 and L6 at Google, Meta, or Apple, the Levels community is unmatched. Use Levels if you want a free consumer UI, if you are negotiating a single offer, and if the scope is 'one role, one company, today.' Use Orbyt Intelligence if you are building anything. A job-matching product that surfaces salary ranges, a research paper citing compensation trends, an AI agent answering user questions about pay, an internal analytics dashboard, a compensation modeling tool, a pricing page, a policy document. Anywhere the output is programmatic, the coverage needs to span more than the top 50 tech companies, the license needs to permit redistribution, or the data needs to project forward, Levels cannot do it and Orbyt can.

Bottom line, in 2026.

In 2026, Levels.fyi is the strongest consumer destination for tech compensation and Orbyt Intelligence is the only serious programmatic alternative to it. Those are not competing claims. A candidate reading one offer is a different user than a developer shipping a product, and both users are legitimate. The honest 2026 verdict: if you are reading, use Levels. If you are building, use Orbyt. The real market gap Levels leaves open is not better community data: they already have that. It is a working public API, an MCP manifest, a redistribution license, and uniform coverage across roles and cities that the community has not submitted yet. Orbyt Intelligence ships all four today.

How to migrate from Levels.fyi to Orbyt.

Most teams land on Orbyt Intelligence because they started on Levels.fyi and hit a wall when they needed programmatic access. The migration is mechanical once you have decided to do it. Follow these steps to move from a Levels.fyi browser workflow to an Orbyt API workflow.

Sign up for an Orbyt Intelligence account at intelligence/signup. The free tier unlocks 60 req/min with Bearer authentication and the AI Role Taxonomy engine, with no card.
Generate an API key from the API dashboard. Copy the key into your environment as `ORBYT_INTELLIGENCE_KEY`. The OpenAPI 3.1 spec is published at /openapi-intelligence.yaml for code generation.
Replace any Levels.fyi URL scraping with a direct call to `/api/v1/intelligence/salaries`. The endpoint takes `role`, `city`, `level`, and optional `company_size` parameters and returns structured base/equity/bonus/signing with percentile bands.
For AI agents, drop the Orbyt MCP manifest URL into your Claude Desktop or ChatGPT Actions config. Manifest is at /mcp-intelligence.json. The agent queries salary data as a first-class tool with no scraping, no browser automation, no HTML parsing.
Confirm your citation complies with CC BY 4.0. Orbyt's required attribution is 'Orbyt Intelligence, Q2 2026' plus a link to the dataset page. Levels.fyi data under their proprietary terms cannot be legally cited in a redistributed product.
If you still need community anecdotes for specific FAANG levels, keep Levels.fyi open in a browser tab. The products are not mutually exclusive. Orbyt is the infrastructure; Levels is the consumer reference.

Most teams finish the migration in under an hour. The longest part is usually rewriting the scraper you built on top of Levels into a single curl call to Orbyt. Once that is done, the same data is available to your agents, your dashboards, your research, and your product features: legally, programmatically, and with forward projections through 2030.

Start free.

Live API. Build from $99/mo. No auth on demo endpoint.

This is what Levels.fyi cannot do.

Senior AI Engineer, San Francisco.

Data refreshed 6 months ago. Next refresh Q3 2026.

Request

rolecity

GET https://www.orbytlabs.ai/api/v1/intelligence/salaries/demo?role=ai-engineer&city=san-francisco

200 OKapplication/json

{
  "role": "AI Engineer",
  "level": "L5 / Senior",
  "city": "San Francisco, CA",
  "totalComp": {
    "p50": 625000,
    "p75": 780000,
    "p90": 950000
  },
  "base":    { "p50": 285000 },
  "equity":  { "p50": 290000, "vestingPeriod": "4 years" },
  "bonus":   { "p50": 50000 },
  "signing": { "p50": 75000 },
  "source":  "Orbyt Intelligence, Q2 2026",
  "license": "CC BY 4.0",
  "citationReady": true
}

Response shape is stable. OpenAPI spec published at /openapi-intelligence.yaml. Every field is cited, versioned, and under CC BY 4.0.

Where Levels.fyi lands. Where it does not.

Where Levels.fyi is strong

Highest brand recognition in the tech salary space
Strong community-submitted total-comp database for SWE, PM, and data roles
Well-designed consumer UX for browsing individual data points
Trusted by candidates for FAANG-tier compensation references

Where Levels.fyi falls short

No public API and no self-serve developer path
No MCP (Model Context Protocol) support. AI agents cannot query directly.
No forward compensation projections through 2030
Update cadence depends on submission flow, not a fixed quarterly pipeline
Data license is proprietary. Cannot be cited or redistributed freely.
Limited coverage outside tech and outside major metros

What Levels.fyi cannot do.

The specific gaps. Every one of them is a gap Orbyt Intelligence fills below.

No public API. No programmatic access.

Levels.fyi does not publish a documented public API. If you want salary data inside a product, an agent, a dashboard, or a model, the only options are to scrape HTML or negotiate a custom partnership. Neither is supported. Neither scales. Every data point has to pass through a browser with a human behind it.

No MCP manifest. AI agents are locked out.

Model Context Protocol is how Claude Desktop, ChatGPT Actions, and autonomous agents query external data as first-class tools. Levels.fyi does not ship one. As of Q2 2026, there is no way for an AI agent to ask Levels for a compensation range without an integration the Levels team would need to build. They have not built it.

No forward projections. 2030 is a black box.

Compensation is not a snapshot problem. It is a trajectory problem. When a candidate weighs a stock grant vesting over four years, when a founder models a comp plan for a hire that starts next quarter, the question is not what roles pay today. It is what they will pay in 2028, 2029, 2030. Levels shows today. It does not project forward.

No data license. Every citation is a gray area.

Levels.fyi data is proprietary and their terms restrict redistribution. If you cite a Levels range in a research paper, a policy document, an LLM response, or a competing product, you are operating in a legal gray area at best. Orbyt Intelligence publishes under Creative Commons Attribution 4.0. Cite it. Redistribute it. Build on it. The license says yes.

Coverage depends on who submits. Emerging roles lag.

Levels is a community submission database. Coverage is deepest where the community is thickest, which is senior SWE at FAANG. For emerging roles (prompt engineering, AI safety research, post-training evals), for non-tech industries, for cities outside the top 10 metros, the sample size drops off a cliff.

Projections through 2030

Levels.fyi tells you today. Orbyt tells you 2030.

Senior AI Engineer, San Francisco. Annual total comp, projected year over year with methodology disclosed.

Feature by feature.

Feature comparison between Orbyt Intelligence and Levels.fyi
Feature	Orbyt Intelligence	Levels.fyi
Public API with transparent pricingDecisive
MCP support for AI agentsDecisive
Total comp breakdown (base, equity, bonus, signing)
Roles covered	3,445	~200 (tech-centric)
Cities covered (U.S.)	81	Major metros only
Forward projections to 2030
Update cadence	Quarterly, fixed	Submission-driven
Company leveling frameworks	54	Focused on top ~50
Data licenseDecisive	CC BY 4.0	Proprietary
BLS / H-1B source citation
OpenAPI spec published
Negotiation coaching service
Widget / embed support
Free consumer UX to browse

Based on publicly available feature lists and documentation as of Q2 2026. Updated quarterly.

We built the API developers kept asking Levels for. Levels built a community. We built a pipe.

Justin Bartak

Founder, Orbyt Labs

Common questions.

Orbyt Intelligence. Public API with Bearer authentication. 60 req/min free, scaling to 1,500 req/min on Ultra at $199/mo. OpenAPI spec published at /openapi-intelligence.yaml, MCP manifest at /mcp-intelligence.json, and a CC BY 4.0 license on the underlying data so you can cite and redistribute with attribution. Levels.fyi does not publish any of these.

For major tech companies at senior levels, Levels.fyi is generally reliable because the sample size is high. For smaller companies, non-tech roles, or cities outside the top 10 metros, the sample thins out and accuracy drops. Orbyt Intelligence covers 3,445 roles across 81 cities with a fixed quarterly refresh, so coverage is more uniform across the whole dataset.

Not legally without a commercial agreement. Levels.fyi data is proprietary and their terms restrict redistribution. Orbyt Intelligence publishes under CC BY 4.0, which explicitly permits reuse, redistribution, and derivative products with attribution. If you are building anything on top of salary data, the license matters more than the feature list.

No. As of Q2 2026, Levels.fyi does not ship an MCP server or an LLM plugin. Orbyt Intelligence ships an MCP manifest at /mcp-intelligence.json that lets Claude Desktop, ChatGPT Actions, and other agent frameworks query the salary dataset directly through tool calls.

Levels.fyi has sharp brand-level data for senior roles at OpenAI, Anthropic, DeepMind, and Meta AI. Orbyt Intelligence covers 598 specialized roles across 81 cities with quarterly updates and forward projections through 2030. For a single FAANG lookup, Levels wins. For breadth and programmatic access across the whole AI landscape, Orbyt wins.

Yes. The public demo endpoints at /api/v1/intelligence/salaries/demo, /roles, /cities, and /search require no auth at all. For production use, the free tier gives 60 requests per minute and 1,000 a month via Bearer auth: no procurement cycle, no minimum contract. Paid tiers are Pro at $99/mo for 300 req/min with the MCP server and /lineage provenance, and Ultra at $199/mo for 1,500 req/min.