StellarBase
StellarCloud + StellarGate

Pay only for what you use.
No tiers. No minimums.

One price per capability for both APIs — inference models on StellarCloud and privacy anonymization on StellarGate. Transparent per-unit billing. Start with €10 free credits — no credit card required.

Platform pricing
Self-hosted available

Run the entire stack inside your own infrastructure.

Everything on this page — StellarCloud models and the StellarGate proxy — deploys as Docker / Helm inside your own data centre. Air-gap capable. Per-token pricing becomes an annual licence, unlimited requests. Ideal for regulated industries, public sector, and IP-sensitive workloads.

€10 free credits No credit card required
EUR pricing EU billing, EU VAT
50 req/sec default Higher on request
EU-hosted Zero data retention
StellarCloud · Inference APIs

Open-source models, hosted for you

Each capability priced by the unit that makes sense — tokens, characters, pages, or images.

Capability
Model
Unit
Price
Text Processing
Language Detection Identify the language of a text across 1,000+ languages — including rare and low-resource ones.
GlotLID v3
per 1M requests
€0.02
Lemmatization Reduce words to their base forms. Neural lemmatizer supporting 60+ languages for search and indexing.
Stanza
per 1K documents
€0.03
Named Entity Recognition Zero-shot NER — detect any entity type (person, org, amount, date, custom) without fine-tuning.
GLiNER-relex large
per 1M tokens
€0.02
Entity Linking Resolve detected entities to canonical identifiers in Wikidata or your custom knowledge base.
GLiNER Linker large
per 1K documents
€0.006
Document Processing
OCR One endpoint: detects structure, extracts body text, handles complex tables, and preserves math formulas. Returns clean, structured output from any PDF or image.
StellarOCR
per 1K pages
€0.90
Embeddings & Retrieval
Text Embeddings — Small Fast multilingual dense embeddings for semantic search, clustering, and similarity. 1024-dim.
Qwen3-Embedding 0.6B
per 1M tokens
€0.01
Text Embeddings — Large Higher-quality multilingual embeddings for retrieval workloads where recall and accuracy matter most.
Qwen3-Embedding 8B
per 1M tokens
€0.15
Text Embeddings — Long Context Multilingual embeddings with 8K context. Ideal for whole-document retrieval and long passages.
BGE-M3
per 1M tokens
€0.036
Image Embeddings Visual embeddings for image search and cross-modal retrieval. Self-supervised, no captions needed.
DINOv3 ViT-L
per 1K images
€0.01
Reranking Cross-encoder re-ranking of search results by relevance. One query + up to 100 docs per call.
Qwen3-Reranker 0.6B
per 1K searches
€0.05
Large Language Models · separate input & output pricing
Model · specs
License
Input / 1M
Output / 1M
GPT-OSS 120B 120B 128K ctx
Open-weights LLM by OpenAI. Instruction-tuned, broad capabilities.
Apache 2.0
€0.20
€0.80
Devstral 2 24B 128K ctx
Mistral coding model. Strong on software tasks, tool use, agentic workflows.
Apache 2.0
€0.50
€2.00
Qwen 3.5 397B A17B 397B · 17B active 256K ctx
Alibaba MoE flagship. Multilingual reasoning, long context, frontier-tier quality.
Apache 2.0
€0.70
€3.80
StellarGate · Privacy Proxy

Anonymize any LLM request

Per-token billing on the anonymization engine — same rate across Mode 1 (we proxy to the LLM) and Mode 2 (you call the LLM). In Mode 1 the LLM provider cost is passed through at their public rate with no markup. Mode 3 is a self-hosted annual license. HITL is an add-on.

Capability
Unit
Price
Transparent Proxy Mode 1
One endpoint. We anonymize, forward to the LLM provider, and de-anonymize in one round-trip.
per 1M tokens
€0.10 + LLM provider cost passed through at rate (no markup)
Tokenized Handoff Mode 2
We anonymize and return tokens. You call the LLM directly. We de-anonymize the response.
per 1M tokens
€0.10 you pay your LLM provider separately
Self-Hosted Mode 3
Unlimited requests inside your own infrastructure. Docker or Helm. Volume-tiered annual license.
annual license
Contact
HITL approval Add-on
Human-in-the-loop approval before sanitized payload leaves the building. Per-approval fee, or €15 / reviewer / month for unlimited.
per approval
€0.05
Included with every request: 15+ entity categories auto-detected, custom dictionaries + regex, per-model policies, dry-run testing, and audit logging. Zero data retention, EU-only hosting.

Volume discounts from 15%

Monthly spend above €1,000? Automatic tier discounts. Above €5,000? Let's talk custom pricing.

Common questions

Do I need a credit card to start?

No. Every new account gets €10 in free credits — shared across StellarCloud inference and StellarGate anonymization. Enough to test everything before committing.

How am I billed?

Pay-as-you-go with monthly invoicing. StellarCloud and StellarGate usage are aggregated on one invoice, charged on the 1st of each month. Pre-paid credits also available.

How is StellarGate billed?

Per-token on the anonymization engine — €0.10 per 1M tokens, covering both directions (anonymize + de-anonymize). Mode 1 and Mode 2 use this same rate. In Mode 1 the LLM provider cost is passed through to your invoice at the provider's public rate, with no markup from us. Mode 3 (self-hosted) is an annual license with unlimited requests.

Is my data used to train models?

Never. Zero data retention across both products. We don't log, store, or train on your prompts, responses, or anonymization tokens. EU-hosted, GDPR compliant.

Can I set budget limits?

Yes. Hard spending caps and real-time usage alerts per-account and per-API-key, across both StellarCloud and StellarGate.

Do you charge in EUR or USD?

All pricing and billing is in EUR. No FX conversion for EU customers.

Start free

€10 free credits. No credit card.

Sign up, generate an API key, start building in minutes.

Back to StellarCloud