All plans are metered in ECU (Euriklis Computation Units) — a single unit that normalises compute cost across tensor operations, graph algorithms, and econometric models. Free and Researcher run on managed CPU capacity; Pro adds higher-throughput CPU execution; Enterprise adds dedicated GPU-accelerated infrastructure.
Full access to the hosted compute API and MCP server, no credit card required.
More ECU for data scientists and researchers who need extended CPU computation.
We are currently rolling out our API and MCP tools. Paid plans will be available very soon — register for free to get early access.
Higher-throughput CPU execution for compute-intensive workloads.
We are currently rolling out our API and MCP tools. Paid plans will be available very soon — register for free to get early access.
Dedicated GPU-accelerated infrastructure for large-scale workloads on a vendor-agnostic backend.
An Euriklis Computation Unit (ECU) is a normalised measure of compute work. A simple vector addition costs a few ECU; a large matrix multiply or a PageRank run on a dense graph costs orders of magnitude more. ECU lets you reason about quota in one number regardless of which operation or library you call.
Free and Researcher run on managed CPU capacity. Pro adds higher-throughput CPU execution across x86_64 and ARM64 architectures, delivering lower latency for compute-intensive workloads. Enterprise adds dedicated GPU-accelerated infrastructure on a vendor-agnostic backend. The API surface and result semantics are identical across all tiers.
No. The Euriklis platform is proprietary and is not distributed in source form. All customers — Free, Researcher, Pro, and Enterprise — consume the platform through the hosted HTTP API and the Model Context Protocol server, with identical result semantics across tiers.
Custom plans are negotiated annually and include dedicated GPU capacity, a production SLA, and volume pricing. Contact us to discuss your workload requirements.