Compute
Storage
For consumers
Workloads
Why Hivenet
Sign in to Hivenet
Access your files and manage your storage account.
Store
Access your files and manage your storage account.
Compute
Launch instances and manage your compute resources.
Store and Compute still use separate accounts today. We’re working on a unified sign-in.
Just need to send a file?
Use Send directlyNew to Hivenet? Get Started
Get started with Hivenet
Choose what you want to use first.
Store
Back up photos and files across your devices.
Compute
Launch self-serve GPU compute for demanding work.
Store and Compute still use separate accounts today. We’re working on a unified account experience.
Need Hivenet for business?
Talk to salesNeed to send a file?
Open SendAlready have an account? Sign in
Contact Hivenet
Choose the path that fits your question.
General contact
Questions, support, partnerships, press, or anything else not tied to a sales conversation.
Talk to sales
Ask about Compute, storage, private AI, or a larger business deployment.
Need help with setup, billing, or troubleshooting? Start with Support for the fastest answer.
Compute - Inference API
Run open-source and foundational models through a managed inference API.
Deploy dedicated endpoints without operating the serving stack yourself. Hivenet Inference API gives teams OpenAI-compatible endpoints, per-replica pricing, and regional deployment paths for production AI workloads on Policloud-backed infrastructure.
OpenAI-compatible API
Dedicated endpoints
One router per client
One region per endpoint
Per-replica pricing
Billed by the second
France, UAE, and US deployment paths
Qwen · Llama · Mistral · Falcon · GPT OSS · Gemma
Managed inference, in your jurisdiction, with RTX economics.
Teams running production AI workloads usually face three trade-offs: raw GPU rental gives maximum control but leaves your team running the router, gateway, scaling, authentication, and observability; traditional sovereign clouds keep workloads closer to the right jurisdiction but often price around data-center GPU infrastructure and commercial commitments; US managed APIs are easy to integrate but can move data and audit trails outside the jurisdiction your customers care about.
Hivenet Inference API gives you another path.
Use managed endpoints for open-source and foundational models, pinned to a selected region, with Hivenet operating the router, gateway, replicas, observability, and endpoint layer.
The API experience stays familiar.
Use an OpenAI-compatible endpoint, update your base URL, and keep common client patterns across OpenAI SDK, LangChain, LlamaIndex, and curl workflows.
The cost model is dedicated and predictable.
Use per-replica pricing billed by the second for steady production workloads where a known infrastructure cost is easier to manage than a token meter.
For teams running production LLM workloads at real volume.
Hivenet Inference API is for teams that want the cost and control benefits of open-source and foundational models, but do not want to manage GPUs, serving engines, replicas, monitoring, and endpoint reliability themselves.
Production teams using LLM APIs
Test Qwen model classes against your prompts, latency target, context length, and cost-performance needs. Smaller Qwen models can fit efficient GPU workflows, while larger classes need testing before production.
SMBs with growing AI bills
Run distilled DeepSeek workloads where the model size fits RTX 4090 or RTX 5090 hardware. Treat larger reasoning workloads as benchmark candidates before production use.
Teams with residency needs
Choose the jurisdiction where your endpoint runs, with deployment paths across France, the UAE, and the US.
Developers using OpenAI-compatible tooling
Keep familiar client patterns. Change the base URL, use Bearer authentication, and keep moving.
Managed endpoint or compute instance?
Use Hivenet Inference API when you want a managed endpoint. Use Compute with Hivenet when you want GPU or CPU infrastructure and full control over the stack.
Need
Use
Why
I want an OpenAI-compatible endpoint
Hivenet Inference API
Hivenet operates the serving layer
I want predictable dedicated inference capacity
Hivenet Inference API
Per-replica pricing helps steady production workloads
I want to run vLLM, TGI, SGLang, llama.cpp, or PyTorch myself
GPU/CPU rental with Hivenet
You control the instance and the stack
I want to fine-tune, test, or run custom pipelines
GPU/CPU rental with Hivenet
Raw GPU/CPU instances give more environment control
I want a custom AI system on sensitive data
Private AI
Hivenet can help scope model, data, deployment, and support needs
Swipe left to see more
Built for the AI workloads SMBs actually run.
Open-source and foundational models are a strong fit for many production tasks when the model is matched to the workload and tested against real data.
RAG
Serve retrieval-augmented generation workflows for internal knowledge, customer support, documentation, and business data.
Structured extraction
Extract dates, entities, categories, and structured fields from documents, messages, tickets, invoices, or records.
Summarization
Summarize documents, conversations, support threads, research, and operational content.
Classification
Classify messages, records, tickets, documents, and workflows using dedicated endpoints.
Code assistance
Support code-related workflows where model quality, cost, and data placement fit the job.
Internal tools
Build AI features for teams that need predictable cost and clearer control over where inference runs.
What teams actually run on Compute.
Open-source and foundational models are a strong fit for many production tasks when the model is matched to the workload and tested against real data.
Choose a model
Select from a managed catalog of model families such as Qwen, Llama, Mistral, Falcon, GPT OSS, and Gemma.
Pick the right tier
Match the endpoint tier to the model size, throughput target, latency needs, and replica count.
Choose a location
Pin the endpoint to an available jurisdiction, such as France, the UAE, or the US.
Swap the base URL
Use an OpenAI-compatible API surface so existing OpenAI SDK, LangChain, LlamaIndex, LiteLLM-style, or curl workflows can connect with minimal changes.
Monitor and adjust
Review endpoint metrics such as requests, tokens, latency, errors, cost, and time to first token where available.
Dedicated endpoint proof, with the methodology behind it.
Each benchmark shows what was tested: model, precision, replica tier, GPU configuration, request rate, latency, and region.
Endpoint uptime
99.7%
Select from a managed catalog of model families such as Qwen, Llama, Mistral, Falcon, GPT OSS, and Gemma.
Qwen baseline
P95 under 30s
Match the endpoint tier to the model size, throughput target, latency needs, and replica count.
Time to first token
Under 5s
Pin the endpoint to an available jurisdiction, such as France, the UAE, or the US.
Integration
OpenAI SDK compatible
Use an OpenAI-compatible API surface so existing OpenAI SDK, LangChain, LlamaIndex, LiteLLM-style, or curl workflows can connect with minimal changes.
Routing
Single-tenant
Review endpoint metrics such as requests, tokens, latency, errors, cost, and time to first token where available.
Model families for production workloads.
The launch catalog focuses on practical model classes for SMB workloads: small and medium models for cost-efficient throughput, and larger models for higher-quality tasks where the workload justifies the replica setup.
Model family
Typical use
Endpoint path
Qwen
Extraction, RAG, structured output, multilingual tasks
Managed endpoint or Compute path
Llama
RAG, summarization, assistants, internal tools
Managed endpoint or Compute path
Mistral
Instruction-following, summarization, tools, European workloads
Managed endpoint or Compute path
Falcon
Smaller and efficient inference workloads
Managed endpoint
GPT OSS
General model inference workloads
Managed endpoint or Compute path
Gemma
Smaller model workflows and experiments
Managed endpoint or Compute path
Swipe left to see more
Qwen for production extraction and RAG
Qwen is a strong starting point for teams testing structured extraction, RAG, and production workflow automation.
Llama and Mistral workloads
Run widely adopted model families for RAG, summarization, internal tools, and model-serving experiments.
Managed catalog to start
Hivenet Inference API starts with a managed catalog. If you need custom weights, fine-tuned models, or a private deployment path, talk to sales.
Predictable pricing for dedicated endpoints.
Hivenet Inference API uses per-replica pricing billed by the second. That gives production teams a clearer way to budget steady workloads than a token meter that grows unpredictably with usage.
Per-replica pricing
Pay for the endpoint capacity you deploy, with pricing tied to the model tier and replica setup.
Billed by the second
Billing follows actual runtime instead of forcing every workload into monthly commitment logic.
EUR pricing
Plan inference spend in EUR where supported, with USD pricing available for US-facing plans.
No token meter for dedicated endpoints
Dedicated endpoints are built for production teams that want a known infrastructure cost.
Tier
Example use
Price
Small endpoint
Smaller models and efficient workloads
from €0.62/hr
Fast small endpoint
Smaller models with more throughput headroom
from €1.10/hr
Medium endpoint
Mid-sized models such as Qwen-class workloads
from €2.10/hr
Large endpoint
70B-class workloads where supported
from €3.80/hr
Swipe left to see more
Enterprise-grade infrastructure for regional AI endpoints.
Hivenet Inference API runs on Policloud-backed infrastructure designed for workloads that need predictable performance, clear regional placement, and a trusted infrastructure path. The point is not hardware ownership as a claim. The point is reliable infrastructure your team can explain.
Single-region endpoint
Pick the region at deploy time. The endpoint stays tied to that region.
Single-tenant routing
One router per customer helps avoid noisy-neighbor routing patterns and keeps the inference path easier to explain.
Full-stack operation
Hivenet operates the router, gateway, runtime, and billing layer for the managed endpoint.
Infrastructure path you can explain
Run inference on a Policloud-backed path instead of routing production AI entirely through default hyperscaler APIs.
Test quality and throughput on your real workload.
Model quality is workload-dependent. Hivenet Inference API is strongest when the model meets your quality bar and dedicated capacity improves cost, throughput, or residency compared with your current API path.
Middleware-tuned throughput
Built for performance per euro
Hivenet's inference layer is designed to improve throughput per euro on Hivenet hardware. Benchmark results show the workload, model, hardware, and endpoint configuration behind each number.
Model evals
Your prompts decide the fit
Test the model against your real traffic, quality requirements, latency targets, and output format before moving production volume.
Endpoint metrics
Watch the signals that matter
Track requests, tokens, latency, errors, cost, and time to first token where available.
Keep your client code familiar.
Hivenet Inference API is OpenAI-compatible, so teams can keep common client patterns and update the endpoint configuration instead of rewriting the integration from scratch.
Built for performance per euro
Hivenet's inference layer is designed to improve throughput per euro on Hivenet hardware. Benchmark results show the workload, model, hardware, and endpoint configuration behind each number.
Your prompts decide the fit
Test the model against your real traffic, quality requirements, latency targets, and output format before moving production volume.
Watch the signals that matter
Track requests, tokens, latency, errors, cost, and time to first token where available.
# Use the OpenAI client, pointed at Hivenet
from openai import OpenAI
client = OpenAI(
api_key="HIVENET_API_KEY",
base_url="https://api.hivenet.example/v1"
)
response = client.chat.completions.create(
model="qwen-example",
messages=[{"role": "user", "content": "Summarize this document."}]
)
Built around real production needs.
Hivenet Inference API is being shaped around teams running high-volume production AI workflows: document automation, extraction, RAG, support workflows, and internal tools.
Example production profile
High-volume document automation
A business automation team uses a dedicated Qwen endpoint in France for part of a production extraction workflow.
Example buyer
SMBs with €5K–€50K monthly API spend
The strongest fit is a team already spending meaningful money on production LLM APIs and looking for predictable dedicated capacity.
Need a different AI infrastructure path?
Hivenet Inference API is the managed endpoint path. Other workloads may be better served by GPU/CPU rental, Private AI, RAG, or S3 storage.
GPU/CPU rental
Rent RTX 4090, RTX 5090, or vCPU instances when your team wants full control over the instance, framework, and serving stack.
Private AI
Work with Hivenet on guided AI projects involving sensitive data, model choice, deployment support, or custom requirements.
RAG
Build retrieval systems on your own data using Hivenet's AI and storage paths.
S3-compatible storage
Store datasets, documents, and AI pipeline artifacts with S3-compatible tools and free egress.
FAQ
Common questions
Move one production AI workload to Hivenet.
Bring your current API usage, model needs, latency target, region requirements, and quality bar. We'll help you decide whether a managed foundational model endpoint is the right path.
