Turn any compute resource
into deployable, trustworthy AI Tokens — fast
Any GPU — a card at home, a traditional data center, a GPU server or a small node — can deploy models and join Krixvon in minutes, turning idle or under-used compute into callable, sellable AI Token services. Text and images are supplied side by side, with video generation on the roadmap. Compatible with OpenAI and New API, backed by a trust layer and fairness mechanisms so this compute can be used with confidence.
18+
Providers connected
6
Routing strategies
1
OpenAI-compatible gateway
Per-token
Usage billing granularity
Three roles, one channel
Providers supply, the platform matches and routes, customers consume the API — a three-tier architecture that makes AI supply flow and bill like a utility.
Provider
Providers
Bring API sources and cost pricing from OpenAI / Claude / Gemini / DeepSeek / local GPUs and more.
Krixvon Platform
Aggregation platform
Aggregate, classify, route, govern, bill and monitor — a single unified supply to the outside.
Customer
Downstream customers
API aggregators, AI SaaS and enterprises call directly with a single platform key.
One request, the best supply found automatically
Downstream sends a request via the familiar /v1/chat/completions interface; the platform picks the best-fit supply-source model by routing strategy and fully logs latency, tokens and cost.
Unified OpenAI-compatible interface, zero learning curve to integrate
Six routing strategies plus automatic failover
Every request logs usage and margin — transparent billing
from openai import OpenAI
client = OpenAI(api_key="$KRIXVON_KEY", base_url="https://foundry-gw.krixvon.net/v1")
r = client.chat.completions.create(
model="claude-opus-4.8",
messages=[{"role": "user", "content": "Hello"}],
)
print(r.choices[0].message.content)Available models (partial) — click to switch
More than text — images and video, one channel
Most aggregators only handle text tokens. Krixvon lets providers supply both text and generative imagery from the same GPU, so customers go from chat to image in one place with a single key and a single wallet.
On-prem generation flow · your GPU turns prompts into work
cinematic lighting
Customer sends a single prompt
ComfyUI · diffusion denoising
Local compute resolves the image step by step
Images render in real time, billed per image; the same flow will power video generation — one key for the customer, from chat to image to clip in one place.
Text · Chat
OpenAI / Claude / Gemini / DeepSeek and local LLMs, billed per token.
Image generation
ComfyUI nodes supply cleanly-licensed models like Z-Image / FLUX / Qwen-Image, billed per image.
Video generation
Open-source video models like Wan2.2 / LTX, billed per second — infrastructure in planning.
One wallet × per-modality metering
One account balance: text priced per token, images per image, video per second; provider revenue share floats with quality. Cleanly-licensed, commercially-usable open-source models go live in one click from the Model Marketplace based on machine specs.
One command, live in minutes
Any machine anywhere with a GPU and outbound internet can join as supply — no public IP, port forwarding or firewall setup required.
▋
// Required
- Outbound connectivity (platform / Cloudflare / model sources)
- One GPU + a supported engine (vLLM / Ollama …)
- Register as a Provider and get a connection token
// Not required
- A static public IP
- Port forwarding / firewall setup
- Configuring Cloudflare yourself (the tunnel is created automatically)
Once connected, traffic is allocated by trust tier (community / verified / contracted data center) and quality routing; users can require trusted nodes only.
Aggregate · Route · Bill · Monitor
Aggregate
Supply aggregation
Bring AI / token / GPU inference APIs from many providers under one roof for centralized classification and control.
Route
Smart routing
Six strategies — lowest cost, lowest latency, highest stability, weighted, round-robin and priority — with automatic failover.
Bill
Usage billing
Log tokens per request, bill customers, reconcile supply cost and platform margin, and deduct balance in real time.
Monitor
Supply monitoring
Continuously measure latency and error rates, auto-degrade on anomalies, and see the whole supply network at a glance.
Built for API platforms, AI SaaS and enterprises
API aggregators
Downstream platforms like New API and Mix Router reach many supply sources with a single key.
AI SaaS teams
Integrate fast via the OpenAI-compatible interface — focus on product, skip building supply and billing yourself.
Enterprise teams
Centrally govern multi-model usage, cost and permissions to meet internal-control and audit requirements.
Providers
List your own APIs or local compute and let the platform handle routing, billing and settlement.
Platform highlights
Provider Management
Self-service onboarding, automated validation, AES-256 key encryption, and centralized management of many supply sources.
Channel Routing
Wrap models from multiple supply sources into priceable, sellable external Channels, routed automatically by strategy.
Usage-based Billing
Token-level billing and margin accounting, customer balance management, and provider revenue settlement — fully traceable.
Multimodal Generation
Text and images supplied and billed through one channel (video on the roadmap) — one wallet, per-modality metering, one key for the customer.
Model Catalog + Probe
Nodes auto-report GPU specs on connect; the platform recommends cleanly-licensed models that fit, installable and live in one click.
Playground
A built-in panel for live availability checks and multimodal API testing — verify each model actually works.
Local GPU Supply
Supports Ollama / vLLM / LM Studio text inference and ComfyUI image generation — multiple engines coexisting on one machine.
Security & Isolation
API keys stored as hashes only, supply keys encrypted at rest, prompt content never logged — secure by default.
Ready to take your
compute to market?
Whether you're a Provider sharing idle GPUs or a developer seeking lower-cost AI APIs.
Provider
Turn idle GPUs into revenue
List your own APIs or local Ollama / vLLM compute and let the platform route, bill and settle — even idle compute keeps earning.
List your compute nowDeveloper
Access AI APIs at lower cost
Reach many supply sources with a single OpenAI-compatible key; the platform picks the most cost-efficient channel, bills by usage, and adds zero infrastructure burden.
Start using the APIReady to unify your AI API supply?
Whether you're a downstream platform tapping many supply sources or a provider monetizing your own APIs and compute, Krixvon handles routing, billing and monitoring for you.