Cost per user
flat rate per API key
The problemPer-token pricing ties your unit economics to usage you can’t fully control.
HelmcodeA fixed monthly rate per key: predictable cost per active user, no matter how heavy they are.
industries · AI-native products
Scale users, not your inference bill.
Cost per active user decides whether the business model holds as you grow. A flat rate per API key, not per token, keeps the line fixed while you scale.
the economics
The economics, by design.
Each cost lever mapped to a platform capability. Sovereignty comes by default.
Unlimited tokens
no consumption caps
The problemUsage-based bills punish exactly the engaged users you want.
HelmcodeRate limits are per key (RPM, concurrency), not on total tokens: a single key moves hundreds of millions a month.
Sovereignty
EU · zero logs
The problemUser data flowing through a foreign API is a liability as you grow.
HelmcodeEU-only inference with zero logs, a clean privacy story for your own customers.
This page is an informational overview, not legal advice. For your obligations and the risk classification of each system, consult qualified legal counsel. AI Act Guide →
what shipping a model makes you
When you become the model’s provider.
An AI-native company does not merely use models, it ships them inside a product. The AI Act has a chapter for that, and it decides which obligations are yours rather than upstream. Open weights change the answer in a specific, written-down way.
01
Four duties if you provide the model
Article 53 asks a general-purpose model provider for technical documentation per Annex XI, documentation for the downstream providers who build on it, a policy to comply with Union copyright law including reservation of rights, and a sufficiently detailed public summary of the training content following the Commission template.
02
A free licence drops two of them
Paragraph 2 exempts models released under a free and open-source licence allowing access, use, modification and distribution, with parameters and architecture information made public, from the two documentation duties. Note what survives: the copyright policy and the training-content summary still apply.
03
Systemic risk cancels the exemption
The carve-out does not reach general-purpose models with systemic risk, which keep the full set of obligations however they are licensed. The licence works as a threshold rather than a shortcut, and which side of it you land on depends on the model you picked.
EU AI Act · Regulation (EU) 2024/1689 Regulation (EU) 2024/1689 of 13 June 2024, Article 53, paragraphs 1 and 2, with Annex XI for the technical documentation and Directive (EU) 2019/790 for the copyright reservation. Obligations for general-purpose AI models have applied since 2 August 2025. read the report →
use cases
Your most common use cases.
The cases with the most traction in the sector, each with its own page in detail.
Semantic search & recommendation Semantic search and recommendation as a core product feature, at flat cost. Customer support In-product assistants and support that scale with users, not with the bill. Voice & transcription Voice features with an open pipeline when your product needs them. Synthetic data Demo and test data that looks like a real account and belongs to nobody, generated as fast as you need it. Classification & triage Labelling and routing inside your own product, at the per-user volume that decides your margin.
Recommended open models.
A starting point per task type. The full guide maps 80 cases to the open model for each one.
- DeepSeek V4 Flash in Helmcode
- The flat-rate workhorse: summarization, RAG, extraction at any volume.
- Gemma 4 12B
- Minimum-cost classification and routing, fine-tunable on your data.
- Whisper large-v3 in Helmcode
- Transcription for voice-enabled products, at marginal cost.
We are distilling and quantizing these open models into small, tightly specialised versions, trained for one task rather than for all of them. A model like that runs on less hardware, answers faster and fits where the big one does not, your own datacenter included. If you have a process with volume and stable criteria, that is the conversation we want to have with you.
// faq
Questions, answered.
What the sector's technical, compliance and business teams ask.
What does "unlimited tokens" actually mean?
No caps on total consumption. Rate limits apply per API key (requests per minute, concurrency), not on how many tokens you process, so a single key can handle hundreds of millions of tokens a month.
How does the cost compare to a closed API?
A closed API charges per token, so cost rises with every active user. A flat rate per key fixes it: past a modest volume, the per-user economics are dramatically better, and predictable.
Is it a drop-in migration?
Yes. Change the base URL and key; OpenAI-compatible SDKs and tools keep working unchanged, most teams ship the same day.
Where is user data processed?
Only on EU infrastructure, with zero logs: prompts are never stored and never train models, a clean privacy posture to offer your own users.
// get started
START BURNING TOKENS
Skip the AI infra work. Deploy your first private inference endpoint today.
Flat rate. EU data. OpenAI API compatible.