The European AI inference network

open ai models.
distributed gpu pool.

One OpenAI-compatible API. Open-source models. Served by a verified pool of GPUs across Europe — from gaming rigs to datacenters. Builders pay less. Hardware owners get paid for every token their machines serve.

OpenAI-compatible EU-sovereign option Full NVIDIA & AMD support
ElephantPool — the distributed AI inference network
80% Of token revenue → hardware owners
~½ price vs EU sovereign clouds (70B-class)
100% Open-source & open-weight models
0 Lock-in, for builders and owners
One pool, two sides

How ElephantPool works

Developers bring requests. Hardware owners bring GPUs. Our scheduler, verifier and payment rails sit in the middle — and take 20%, only when tokens flow.

1

Builders call the API

Swap your base URL, keep your code: an OpenAI-compatible endpoint for Llama, Qwen, DeepSeek, gpt-oss. Pay per token, prepaid credits, no subscription.

2

The pool serves it

The scheduler routes each request to the best verified machine — by speed, price, reputation and (if you choose) EU-only residency. Every job produces a cryptographic receipt.

3

Owners get paid

80% of every token's revenue goes to the machine that served it. Monthly SEPA payout via Stripe. Paid for work performed — nothing to buy, nothing to lock up.

80%
20%
→ Hardware owners, per token served → Network: scheduling, verification, payments

The split is public and applies to every job on the network.

For Builders

Models & pricing

The latest open models, transparent per-token pricing, in euros. The EU-Sovereign tier routes only to attested machines physically in the EU — and still undercuts every sovereign cloud.

Model Standard
€ / M tokens in · out
EU-Sovereign
€ / M tokens in · out
Elsewhere
cheapest US · EU sovereign
Llama 3.1 8B
8B dense · 128K
€0.03 · €0.06 €0.05 · €0.10 US  $0.02 · $0.04 DeepInfraEU  —
Qwen3.8 27B
27B dense · 262K · Apache-2.0 · new
€0.25 · €1.50 €0.35 · €2.00 US  $0.40 · $3.00 ChutesEU  —
Gemma 4 26B-A4B
26B MoE · 256K · Apache-2.0
€0.06 · €0.25 €0.10 · €0.35 US  $0.07 · $0.34 DeepInfraEU  —
Llama 3.3 70B
70B dense · 128K
€0.30 · €0.50 €0.35 · €0.55 US  $0.10 · $0.32 DeepInfraEU  €0.65 · €0.65 IONOS
gpt-oss-120b
117B MoE · 131K · Apache-2.0
€0.06 · €0.30 €0.10 · €0.45 US  $0.03 · $0.17 CoreWeaveEU  €0.08 · €0.40 OVHcloud
DeepSeek V4-Flash
304B MoE · 1M · MIT
€0.15 · €0.35 €0.25 · €0.55 US  $0.08 · $0.18 DeepInfraEU  €0.40 · €0.80 Scaleway
MiniMax M3
427B MoE · 1M · community license
€0.25 · €1.00 €0.35 · €1.30 US  $0.28 · $1.10 DeepInfraEU  —
GLM-5.2
753B MoE · 1M · MIT
€0.45 · €1.40 €0.60 · €1.80 US  $0.48 · $1.49 NovitaEU  €1.80 · €5.50 Scaleway
Qwen3.5 397B-A17B
397B MoE · 262K · Apache-2.0
€0.40 · €2.50 €0.50 · €3.00 US  $0.30 · $1.93 DigitalOceanEU  €0.60 · €3.60 Scaleway/OVH

Target launch pricing. Batch tier: −50%. Competitor prices from public price lists, August 2026. Free starter credits on small models for early users.

All catalog models are open-weight (Apache-2.0, MIT or Llama/community licenses — linked per model in the API). Network-scale giants (Qwen3.8 2.4T, Kimi K3, DeepSeek V4-Pro) require multi-node serving and are on the roadmap, not the launch catalog.

Drop-in replacement

OpenAI-compatible /v1/chat/completions, streaming included. Change one base URL and you're on the pool.

GDPR by construction

Prompts never touch a disk, are never used for training, and the EU-Sovereign tier never leaves attested EU machines. DPA available.

Smart routing, smart savings

Pick a model — or pick auto and let the router send each request to the cheapest model that can handle it. Batch API at half price.

For Hardware Owners

Your GPU has a day job now

Install mahout, our open-source agent — named after the person who guides an elephant. It benchmarks your machine, serves models it can handle, and gets you paid for every token. Full NVIDIA and AMD support, from day one.

Entry

RTX 3090 / 3080

10–24GB VRAM · serves 8–32B models
est. €60–120/mo gross, well-utilized · you pay the power
Estimate
Sweet spot

RTX 4090 / 5090

24–32GB VRAM · serves 32B-class fast
est. €130–290/mo gross, well-utilized · dual-card rigs serve 70B
Estimate
AMD first-class

RX 7900 XT / XTX

20–24GB VRAM · full ROCm & Vulkan support
est. €60–110/mo gross, well-utilized · same features & pay as NVIDIA
Estimate
Pro / datacenter

A6000 · A100 · H100

48–80GB VRAM · serves 70B+ solo
est. €200–700/mo gross per card, well-utilized · SLA tier access
Estimate

Safe by design

mahout is open source. It runs only signed engines in sandboxed containers — read-only weights, zero network egress, never any client code. No inbound ports on your router.

Your machine, your rules

Set availability windows, power limits, disk quota and a price floor. Auto-pause when you game or work. One-click stop, always.

Paid for work, monthly

80% of every token your machine serves, tracked by cryptographic receipts you can verify yourself. Monthly SEPA payout from €50, processed by Stripe. Guaranteed starter work in your first week.

Earnings Estimator

What could your hardware earn?

Estimates for continuously batched serving at our launch price book, with 80% of token revenue paid to you. You are paid for work actually routed to your machine — this is an estimate, not a promise.

Assumptions used: Launch price book, continuous batching Your share: 80% of token revenue Power draw under load per rig shown model-by-model Utilization depends on network demand in your region
Payout rate at full load €0.35/hr
Utilization 50%
Your electricity cost −€42
Monthly net €86
Per year (net) €1,032

⚠️ Estimates only. You are paid for tokens your hardware actually serves; volume depends on network demand, your machine's benchmark, uptime and reputation. This is payment for a service you provide — not an investment, not a yield product.

Credibility & Transparency

Don't trust us. Verify us.

Verified compute

Every job produces an activation fingerprint (the TOPLOC scheme — 258 bytes per 32 tokens). A random sample of all work is replayed on trusted hardware. Machines that fake work don't get paid — and get banned.

Public receipts

Signed job receipts are batched into a Merkle tree and anchored on a public blockchain every 10 minutes. Any host or client can independently prove what was served, when, and what was owed.

Open source, open docs

The mahout agent is open source and auditable. Our architecture, economics and verification design are published: read the whitepaper and the technical spec.

European, by choice

Operated from Belgium by De Buck Technologies. EU-Sovereign tier with attested EU-only routing, zero retention, and a standard DPA — at roughly half the price of sovereign clouds.

One public split: 80 / 20

80% of token revenue to the machine that did the work, 20% to the network. No hidden spreads, no house hardware jumping the queue on equal terms.

Built to compound

Long term, idle pool capacity trains open models collectively — the network that serves AI also improves it, in the open. That roadmap is public too.

Questions

Frequently Asked

Is this safe for my PC?

mahout runs only signed engines in sandboxed containers — no client code, no network egress from the engine, no inbound ports. It's open source, so you don't have to take our word for it. Auto-pause keeps it out of your way when you're using the machine.

How and when do I get paid?

80% of the revenue of every token your machine serves. Monthly SEPA payout from €50, processed by Stripe (which also handles identity verification). Your dashboard shows live, receipt-backed earnings.

What if my internet or PC is slow?

mahout benchmarks your machine and the network only routes work it can genuinely handle — smaller models on smaller rigs. Slow uplinks serve batch jobs instead of live chat. Everything is measured, nothing is promised that your hardware can't do.

Is AMD really fully supported?

Yes — full support, day one. RX 7900-class cards run via ROCm (vLLM) with a Vulkan fallback (llama.cpp): same features, same tiers, same pay as NVIDIA. Our own fleet includes AMD cards, so the AMD path is exercised daily.

Is my prompt data private?

Prompts and completions live in RAM only, are never logged, and are never used for training. The EU-Sovereign tier additionally guarantees attested EU-only machines. A standard DPA is available for businesses.

Why open-source models only?

Open weights are what make a distributed pool possible — anyone can verify what's running, and nobody can take the models away. They're also what makes it cheap: no license margin, just hardware and electricity.

Join the founding cohort

First 100 hosts: priority routing and a founding-host badge. First API users: free starter credits on every model.

Get API Access Connect My GPU

One email. Tell us what you build — or what hardware you have. Response within 24 hours. De Buck Technologies, Belgium.