
LLMBase
Europe-based AI platform, made in Germany
3 followers
Europe-based AI platform, made in Germany
3 followers
European ChatGPT alternative with 80+ AI models — GPT-5, Claude, Gemini, Grok, DeepSeek and open source models in one app. One subscription replaces ChatGPT, Claude and Gemini. GDPR compliant, hosted in Germany.
This is the 2nd launch from LLMBase. View more

LLMBase
Launching today
OpenAI-compatible API for EU-hosted open models, with a free control layer: per-key spend limits, model allowlists, audit log, team budgets. Partner data centres in the EU (Nebius, Hetzner). DPA included. From $0.18/1M input tokens, prepaid from $5.





Free Options
Launch Team

Render Sandboxes: a computer for every agentIsolated compute, persistent state and a real filesystem
Promoted

Hey everyone, I’m Marius, maker of LLMBase.
We first came to Product Hunt in May with a European AI chat. The useful feedback was simple: developers don’t need another chat wrapper. They need a clean way to run strong open models in Europe, with a DPA, without rewriting their stack.
So we rebuilt around that.
LLMBase is now an OpenAI-compatible inference API. You keep your code, point the OpenAI SDK at https://api.llmbase.ai/v1, and add your key. Open models are hosted in the EU, in ISO 27001-certified partner data centres (Nebius, Hetzner). Prepaid from $5, from $0.18 per 1M input tokens. Businesses can sign a DPA in their account. We do not train on your content.
Streaming, tool calls, structured outputs and prompt caching are in the API. There is also a key-management API, so your own systems can issue and limit keys per customer. If you need reserved capacity, dedicated endpoints are available. And if your team just wants to chat, that is still there — Pro starts at $19/month, with a free tier to try it.
On top of the API sits a control layer, free for every account, no Enterprise plan needed. Set monthly and daily spend limits per key, limit each key to the models you approve, and give each team its own budget. Every key change lands in an audit log, usage is reported per key, and only owners and admins can manage keys. A documentation pack exports your DPA status, TOMs, model list and key usage in one file.
What’s new since May:
The inference API is the core product, not an add-on
One OpenAI-compatible base URL for EU-hosted open models
Prepaid token billing, no API subscription
Key management for customer-level keys and limits
A free control layer: model allowlists, audit log, usage per key and team budgets
Dedicated endpoints
A signable DPA in the account
This week is our Black Week. For new customers on Product Hunt: 30% off your first monthly invoice on Pro, Expert or Business, until 1 December 2026, 23:59 Europe/Berlin. Code: BLACKFRIDAY26. Enter the code at checkout — it is not applied automatically from the link. The discount does not apply to Max, Starter, annual plans or API credits, and it does not apply to existing accounts. After the first invoice, the regular price returns.
I’d love your honest feedback, especially if you already use the OpenAI SDK:
Would you switch the base URL for an EU-hosted setup?
What’s missing before you’d put this in production?
Thank you for taking a look.
Marius