News & Press

BasePeak.AI is the AI workspace built for Germany's public sector and regulated SMEs - where data must stay in Germany. Infercom provides the EU sovereign inference layer underneath. Together: a complete AI stack that never leaves German jurisdiction.

German government agencies and regulated businesses face a hard constraint: their data cannot leave Germany. Not just the EU - Germany specifically.

BasePeak built an AI platform around this reality. Their workspace includes chat, document understanding, and agent capabilities - all designed for public sector compliance and German data protection requirements.

Infercom adds the inference layer. SambaNova hardware in our Munich datacenter, processing at up to 10x GPU speeds. No US routing, no CLOUD Act exposure, full EU sovereignty.

For agencies and businesses that need AI but can't compromise on data residency: this partnership delivers both.

"Public sector and SMEs in regulated industries have a strong need for AI systems. But their data must not leave Germany. With our AI platform BasePeak.AI, we deliver that sovereignty. Now with Infercom as inference partner, our customers get even more choice - the fastest open-source models, all EU-hosted."

"German public sector has unique requirements - not just EU, but German data residency. BasePeak understands these constraints. We provide the performance infrastructure. Together: sovereign AI that meets the strictest German standards."

Google's most capable dense open model is now running on EU sovereign infrastructure. Frontier-class reasoning, native multimodal capabilities, configurable thinking mode - all at over 30% faster than the next fastest provider.

Gemma 4 31B brings Google's latest open model research to Infercom's EU sovereign infrastructure:

Frontier-class reasoning: 89.2% on AIME 2026 mathematical reasoning, 84.3% on GPQA Diamond scientific QA. Configurable thinking mode lets you balance deliberation depth against latency.

Native multimodal: Process text and images together for document understanding, chart extraction, and structured data output. Vision-plus-reasoning in a single model.

Agentic capabilities: Native function-calling, structured JSON output, and system-prompt support. Compatible with OpenClaw, CrewAI, and other multi-agent frameworks.

Performance: Over 30% faster than the next fastest provider according to Artificial Analysis benchmarks. Running on SambaNova's dataflow architecture in our Munich datacenter.

Pricing: €0.20 input / €0.35 output per 1M tokens with 128K context window.

CoformAI is an Agentic AI Operating System - describe a use case in plain language and it generates the complete agent stack. Infercom provides the EU sovereign inference underneath. Together, European enterprises get agentic AI that actually ships - with no data leaving Europe.

European enterprises keep asking the same question: how do we run AI agents without data leaving Europe?

This partnership answers it. CoformAI builds the full agent stack from a plain-language description - no ML team, no code, no months of development. Infercom runs it: EU-only inference in Munich, up to 10x faster than GPU alternatives, no US routing.

CoformAI covers every layer an enterprise needs - from how users interact with agents, to multi-agent orchestration, all the way to compliance and governance. Infercom powers the inference layer from the ground up: EU sovereign inference built into the stack, not bolted on.

One partnership. Zero tradeoffs between compliance and shipping.

"CoformAI generates complete AI systems from plain language - no ML team required. We add EU enterprise inference: fast, sovereign, no compromises. Together, European enterprises get agentic AI that actually ships."

"Our customers kept asking: where can we run agents without data leaving Europe? Infercom is the answer. Their EU enterprise inference combined with our Agent OS - a complete stack, fully sovereign."

ABACUS Consulting & Solutions, a Luxembourg IT consulting firm with 14 years serving financial institutions and public sector, joins the Infercom partner ecosystem. Together we're bringing sovereign AI infrastructure to Europe's regulated industries - including LUXIOS, their sovereign AI assistant.

"Europe's regulated industries don't need a revolution in how they work - they need sovereign AI that integrates into the governance and compliance frameworks they've trusted for years; together with Infercom, we're finally giving CISOs a compliant alternative to shadow AI."

"When someone asks 'where does my data go?' - we have a one-word answer: nowhere. It stays in Munich, on EU-owned infrastructure. That's what ABACUS can now offer their clients."

MiniMax M2.7 Ultraspeed, the latest 229B parameter frontier model, is now available on Infercom's EU sovereign infrastructure. 30% coding improvement over M2.5, built-in self-critique, and native multi-agent orchestration - all at 400+ tokens/sec.

M2.7 introduces two major capabilities that weren't possible with previous models:

Built-in Self-Critique: The model automatically reviews and refines its outputs before responding. Better first-attempt results on complex coding and reasoning tasks - no external validation loops needed.

Native Agent Teams: Multi-agent collaboration as a core model capability, not just prompting. Stable role boundaries, adversarial reasoning, and behavioral differentiation built into the model itself.

Benchmarks: 56.22% SWE-Pro (matching GPT-5.3-Codex), 55.6% VIBE-Pro for full project delivery, 57% Terminal Bench 2, 66.6% MLE Bench Lite medal rate.

Works with all major agentic coding tools: Aider, OpenCode, Cline, Cursor, Continue, Goose, Windsurf, and Claude Code.

Pricing: €0.60 / €2.40 per 1M tokens. Running on SambaNova hardware in Munich with full EU data sovereignty.

Infercom now supports the Anthropic Messages API alongside OpenAI-compatible chat/completions and responses endpoints. Use your existing Anthropic SDK code with EU sovereign inference - just change the base URL.

We've added native support for Anthropic's /v1/messages endpoint, complementing our existing OpenAI API compatibility for /v1/chat/completions and /v1/responses.

This means you can now use Infercom with the Anthropic Python or TypeScript SDK directly - no code changes beyond pointing to api.infercom.ai. Tool use, streaming, and system prompts all work as expected.

Both MiniMax M2.7 Ultraspeed and gpt-oss-120b support the Anthropic Messages format. Same models, same EU sovereign infrastructure, more ways to integrate.

Note: Vision, extended thinking, and prompt caching are not currently supported.

Opper users can now route to Infercom for EU sovereign inference - MiniMax-M2.5 and gpt-oss-120b, no code changes required.

Opper is the AI gateway trusted by 50,000+ developers. One API, 300+ models, OpenAI SDK compatible. Change your base URL and you're running on a different provider.

Now Infercom is available through Opper.

MiniMax-M2.5 at 400+ tokens/sec. gpt-oss-120b battle-tested for production. Both running on SambaNova hardware in Munich - full EU jurisdiction, no US CLOUD Act exposure.

For teams in finance, healthcare, legal, and government: your compliance team will thank you. For everyone else: you get up to 10x faster inference than GPU alternatives.

No code changes. Select Infercom in the Opper console and you're live.

"Developers want fast inference and enterprises need data sovereignty. With Opper, they get both - route to Infercom for the fastest EU-hosted inference available, with zero code changes."

"Europe needs more than one sovereign inference provider - it needs a network of them. Every additional EU node shortens latency for developers and keeps the data under European law for enterprises. Infercom raises that bar with SambaNova hardware in Munich, and now our 50,000 developers can reach it through a single API."

EUrouter is an intelligent AI gateway that routes requests to the best provider based on your priorities - speed, cost, or sovereignty. Infercom is now integrated as a first-class provider.

For teams already using EUrouter: you can now route to Infercom for gpt-oss-120b and MiniMax-M2.5 with a single configuration change. Sort by throughput or latency - Infercom ranks at the top for both.

For teams evaluating AI gateways: EUrouter + Infercom gives you intelligent routing with true EU data sovereignty. No US CLOUD Act exposure. No cross-border transfers. Up to 700 tokens/sec on models that stay entirely within European jurisdiction.

This is what the European AI ecosystem should look like - specialized providers working together, each doing what they do best.

DeepMask is an AI workspace for teams who handle sensitive data daily - legal firms, banks, enterprises. Now powered by Infercom's Munich infrastructure for true European data residency.

Your team is already using AI. The question is: where does their data go?

DeepMask gives teams a unified AI workspace with 25+ models, persistent projects, and connectors to Google Drive and SharePoint. Infercom provides the infrastructure - Munich datacenter, SambaNova hardware, no US data routing.

DeepMask built the workspace. We built the inference platform. Now European companies get both - great AI tools and the certainty that everything runs and stays in Europe.

"Deepmask built the workspace. We built the enterprise AI execution platform. Now European companies get both - great AI tools and the certainty that everything runs and stays in Europe."

"We work with legal firms, banks, enterprises who handle sensitive data daily. They don't want 'EU-hosted somewhere.' They want european ownership. Infercom gives us that - fast inference, Munich datacenter, no asterisks."

Cortecs users now get direct access to Infercom's EU-sovereign inference: up to 10x faster on purpose-built AI hardware, hosted in Munich with full GDPR compliance, no US CLOUD Act exposure, and models like MiniMax-M2.5 at 400+ tokens/sec. Intelligent routing meets sovereign infrastructure.

Voice AI needs sub-second response times for natural conversations. European enterprises need data sovereignty. Until now, you had to pick one. Neur-X + Infercom delivers both.

The trade-off that shouldn't have existed: US cloud providers deliver fast voice AI, but with CLOUD Act exposure. EU GPU-based inference keeps data sovereign, but with noticeable lag that customers notice.

Neur-X builds enterprise-grade AI voice agents - inbound call automation, outbound campaigns, and real-time agent assist. Their platform handles everything from telephony to conversation flow, with a fully isolated architecture and no-code configuration.

Infercom provides the sovereign inference underneath. SambaNova's dataflow architecture delivers up to 10x faster inference than GPUs - in our Munich datacenter. Sub-second response times with full EU jurisdiction.

Banking, insurance, healthcare, enterprise support - industries where both speed and sovereignty are required, not optional. Your customers call, the AI responds naturally, and no data ever leaves Europe.

Melious, the sovereign European AI platform, has launched with Infercom as one of its inference providers. Melious users get access to EU sovereign AI inference at 400+ tokens/sec on models like MiniMax M2.5.

Congratulations to the Melious team on going live! Melious is building a sovereign AI platform that orchestrates inference across European providers.

"Even when you use providers like Mistral or Langdock, your data is still hosted on EU servers, but from US companies. There's a problem because of the US Cloud Act," said Dean Fünffrock and Simon Jakobs, co-founders of Melious, on why they're building a fully sovereign European AI alternative.

Great to see the European AI ecosystem growing stronger together.

MiniMax M2.5, one of the world's fastest multimodal models, is now available on Infercom's EU sovereign infrastructure. Independently measured at 400+ tokens/sec by Artificial Analysis, M2.5 delivers strong reasoning, multilingual, and function calling capabilities - all with full EU data sovereignty.

SambaNova's ChipChats series featuring Infercom Managing Director Altug Eker on launching a sovereign AI cloud in 90 days.

Infercom and Co-Mind have joined forces to solve Shadow AI in European enterprises. Co-Mind provides a private AI workspace with chat, agents, and knowledge bases - fully governed with no data leakage. Infercom powers the inference underneath.

Shadow AI is the #1 unaddressed risk in European organizations today. Employees are using ChatGPT, Claude, and Gemini for work - uploading contracts, customer data, and internal documents with no IT oversight, no GDPR compliance, and no control.

Infercom and Co-Mind have built the answer. Co-Mind gives organizations a private AI workspace - chat, agents, knowledge bases, guardrails, and no shadow AI risk. Infercom provides the sovereign inference underneath, running in Germany, fully within EU jurisdiction.

The demand is clear: companies don't want to choose between usability and sovereignty. They want both. And they want the flexibility to run it managed or on-premises. That's exactly what this partnership enables. Your team gets the AI productivity they want. Your organization keeps full control of its data.

Shadow AI ends here.

"If you cannot make it easy enough, people won't adopt. Doesn't matter how sovereign you are."

Infercom's inaugural LinkedIn Live webinar on the strategic landscape of sovereign AI in Europe.

Live AI demonstrations and networking at Soho House Berlin with SambaNova and ecosystem partners.

In-depth Q&A on the SambaNova and Infercom partnership for sovereign AI in Germany.

SambaNova Systems announces strategic partnership with Infercom to deploy Europe's first sovereign AI inference platform in Munich datacenters. The deployment leverages SambaNova's revolutionary dataflow architecture to deliver world-record inference performance with complete EU data sovereignty.

"This partnership represents a watershed moment for European AI independence. Infercom is demonstrating that European organizations don't have to compromise between world-class performance and data sovereignty."