Skip to content
AgentsIndex

Heabsy AI Platform

Verified owner

OpenAI- and Anthropic-compatible inference API from an EU company, with own models on EEA GPUs and no prompt logging.

Best for
Heabsy AI Inference is best for regulated-industry teams who need compatible API access without moving data outside their infrastructure.

What we checked

Pages last read
Facts checked
2 facts

From 3 sources

AI answers
Too few read

58 buyer-question answers read, November 2026

At a glance

Tags
No training on your data, Zero data retention

For developers

Works with
Claude Code, Cursor, Cline

Heabsy AI Platform in our research

All research

Study

AI Tool Pricing Index, November 2026

Pricing

Usage-based

43 of the 361 tools we checked bill by usage.

Edition price, checked 2 November 2026.

Pricing pages of the AI tools, November 2026 editionas of n = 361 tools checked

What is Heabsy AI Platform?

Heabsy AI Inference is an AI inference platform for regulated-industry teams that serves open models through OpenAI-compatible and Anthropic-compatible endpoints. It supports chat completions, SSE streaming, tool calling, structured output, and prefix caching, with per-key budgets and a usage dashboard.

The flagship model runs on dedicated EEA GPUs; the rest of the catalog is routed through global providers and may run outside the EEA. The API is billed per million tokens with no minimum spend; separate fixed-price implementation projects start at €5K (about $5.5K) (Starter) and €25K (about $27K) (Growth), with Enterprise on custom annual contracts.

The platform is a model API alongside a consulting practice

Heabsy presents the Platform as an inference API rather than a general business application. The Platform page also says an Anthropic compatible endpoint is available, so existing agent clients can be pointed at its service with a changed endpoint. The API reference and platform overview describe the interfaces.

This sits beside a separate implementation business. Heabsy describes projects for document assistants, internal search and agents operating in existing systems, while its project page sells discovery and implementation work. The two offers use different purchasing routes and price structures. The company page and project pricing make that split explicit.

Account creation and key management are self-service

The current Platform page calls key setup “Three steps, all self-service.” It says that signing up in the console makes an account ready at once, with no call or contract, then lets a user issue multiple keys with separate budgets and model lists. The public page does not describe this as a sales-assisted onboarding flow. Heabsy’s setup steps are the source for that process.

The API reference says keys are issued and managed in the Platform console, and that requests authenticate with a bearer key. It also documents a 429 response when a key has exceeded its budget or rate limit.

Usage billing and project fees are two different price lists

For API use, the catalog states that pricing is per million tokens, with separate input, cached input and output rates, and says there is no minimum spend or subscription. It groups the available choices behind one key and one invoice from a European counterparty. The catalog is therefore a usage price list for inference, rather than the fixed project price shown elsewhere. The model catalog publishes that billing basis and the individual rate columns.

The consulting page instead lists a Discovery stage from €5,000 (about $5,495) per project and a production oriented Growth stage from €25,000 (about $27,475) per project. It says there is no hourly rate or per user charge, and explains that scope, connected systems, evidence requirements and operating responsibility change the project price.

Model selection changes the data location and retention terms

Heabsy separates models it serves on its own infrastructure from partner hosted choices. Its catalog says prompts and completions are not stored for the former, whereas partner hosted models have no zero data retention guarantee and are not assured to remain in the EEA.

The site gives two qualifications to its European processing language. The Platform page says its own models run in the EEA apart from the two DeepSeek entries it marks as outside the EEA. The privacy policy and DPA say that an organization or key set to the EEA data region is processed only in EEA data centers, while other Heabsy served requests may run in the EEA or United States. The Platform page, privacy policy, and DPA should be read together when location is a requirement.

The published privacy terms separate content from account records

For models on Heabsy infrastructure, the privacy policy says prompt and completion text is processed in memory and not written to durable storage or request logs. It says the company retains usage metadata such as model, timestamp, token counts, status and key identifier for billing, with usage metadata and financial records retained for ten years. Application logs have a stated 30 day retention and security audit logs one year. The privacy policy gives these retention periods.

Heabsy publishes a standard GDPR Article 28 data processing addendum. It identifies the customer as controller for API content, says a signed copy can be requested, and describes audit and inspection rights on reasonable notice. The same document says special category data, government identifiers and children’s data should not be sent without a lawful basis and prior written notice. The DPA sets out those roles, restrictions and audit provisions.

FEYA runs the service from Slovakia for enterprise AI work

The legal notice names FEYA, s.r.o. as the Slovak limited liability company behind the Heabsy brand, identifies Kostiantyn Fomichov as managing director, and describes its business purpose as AI implementation, document assistants, search, agents and language model operation. The public company page gives the same legal entity, registration number and Bratislava address, and says Heabsy is the brand for its implementation services. The legal notice and company facts provide the operator identity.

Heabsy says its primary market is DACH and describes work for regulated industries. Its project materials offer on premise deployment or deployment on its own EEA hardware for a larger implementation. The API is therefore part of a wider enterprise AI practice, while the current Platform page makes a separate self-service route available for inference users. The company page, project page, and Platform page state those respective audiences and routes.

Heabsy AI Platform price and terms

Terms

Free plan or trial
NoNo

For AI agents

Developer docs

Heabsy AI Platform's pricing page

Pricing page, captured

heabsy.com/models, checked

What we checked on Heabsy AI Platform

YesPartlyNo

What it costs to startYes

What the price counts

Priced per use: every model is billed per million input, cached input and output tokens, with no subscription and no minimum spend.

Per-million pricing straight from the gateway's price list — no minimum spend, no subscription.

heabsy.com/modelsChecked

How you build with itYes

Public API

A public API reference documents an OpenAI-compatible endpoint with chat completions, model listing, streaming and tool calls, plus an Anthropic-compatible endpoint for Claude Code.

The API is OpenAI-compatible: point any OpenAI SDK at our base URL and it works.

api.heabsy.com/docsChecked

Heabsy AI Platform alternatives

Questions about Heabsy AI Platform

How much does Heabsy AI Platform cost?
Heabsy AI Platform costs $0.10 per 1m input tokens. From heabsy.com/models, checked 2 November 2026.
Does Heabsy AI Platform have an API or an MCP server?
Heabsy AI Platform has an API. From api.heabsy.com/docs, checked 2 November 2026.
What have we checked on Heabsy AI Platform?
2 facts, under these headings: What it costs to start, how you build with it.
How do we check Heabsy AI Platform?
We hold 2 facts on Heabsy AI Platform, quoted from 3 pages. Facts and price last checked 2 November 2026, when we also found the last change.

How we check every listing

Sources and changes

Last changed

What changed1

  • What it does

Heabsy AI Platform price history

We have read the price once so far, on 2 November 2026: from $0.10 per 1m input tokens.

PricePlanFirst seenLast seenReadingsSource
$0.10 per 1m input tokens2 November 20262 November 20261Pricing page, read 2 November 2026

From our research on 2 November 2026. A date is when we read the page, not when the price changed.

Similar to Heabsy AI Platform

Claimed by the Heabsy AI Platform team.

We check its facts against the Heabsy AI Platform site.Facts checked 2 November 2026

Report an error if a fact is wrong.