Proxied
Claude Code
Routes Claude Code through a local Anthropic-to-Nebius translation proxy. Your subscription, login, and config stay untouched.
nclaude GLM 5.3 Flash is now the default
YOUR AGENTS. OPEN MODELS.
Nebius TF Relay
Use open models
with your existing harness.
Run the coding agents you love on Nebius Token Factory. One local relay. Eight agents. Your setup stays yours.
curl -fsSL https://nebius-tf-relay.vercel.app/install.sh | bashmacOS / LinuxBun is installed automatically if needed
MORE POSSIBILITIES.
01 / GET CONNECTED
Single install.
Multiple uses.
From your terminal to open models in three steps.
No changes to your existing agent configuration.
- 01
Install once
Run the one-liner. It drops
nebiusrelayplusnclaude,ncodex,nopencode,npi, andnprimeonto your PATH and installs Bun if you don't have it. - 02
Add your keys
On first run,
nebiusrelay configureasks for your Nebius Token Factory key and an optional Tavily key for live web search.nebiusrelay configure - 03
Launch an agent
Type
nclaudeorncodexand keep working. The Relay injects Nebius settings for that run only. Nothing is written to your real agent config.ncodex
02 / PICK YOUR AGENT
Familiar tools.
Fresh possibilities.
Use the workflow you already know.
Just give it a different engine.
Proxied
Codex CLI
Talks to Nebius through a local Responses-to-chat proxy, with headless exec support. Sessions stay resumable across providers.
ncodexProvider config
OpenCode
Launches with Nebius wired in as an OpenAI-compatible provider, injected only for that run. Close it and your setup is exactly as it was.
nopencodeProvider config
Pi Code
Starts with a custom Nebius provider and a temporary config directory, while normal local session history keeps persisting.
npiProvider config
Hermes Agent
Nous Research's agent, launched with an isolated home overlay so your sessions and skills stay native while credentials stay ephemeral.
nhermesAlpha
DeepSeek Harness
Boots the DeepSeek web profile with Nebius layered in as a provider. Pairs naturally with DeepSeek V4 Pro and Flash.
ndeepseekProvider config
Grok Build
xAI's terminal harness driving Nebius models. Your key is fenced off from api.x.ai, and the model is told not to claim it is Grok.
ngrokProvider config
Prime Agent
PrimeIntellect's RLM agent, with its persistent IPython tool and subagents running on Nebius models. Your own Prime config stays untouched.
nprimeDESKTOP INTEGRATION / ALPHA
ChatGPT / Codex Desktop
An optional managed profile routes compatible desktop coding tasks through Relay. It changes the shared Codex config until you restore it; it does not replace models in ordinary ChatGPT web chats.
03 / FIND YOUR MODEL
One key.
An open model lineup.
Choose a model for the task at hand.
Switch with a flag. Keep your workflow.
BUILT TO STAY OUT OF YOUR WAY
One relay, eight harnesses
Claude Code, Codex, OpenCode, Pi Code, Prime Agent, Hermes, DeepSeek Harness, and Grok Build all run on Nebius open models through a single local install.
Live web search, built in
The proxy emulates native web_search with Tavily and streams real Anthropic citation blocks straight into your agent.
Cost tracking per session
Session cost estimates use reported token usage and model catalog pricing, with a summary when you exit.
Config-free & self-updating
CLI wrappers use temporary provider settings. Desktop integration is opt-in and persistent. The installed binary checks the release site for updates.
LESS SETUP. MORE BUILDING.
Your next coding session,
powered by Open Models.
Get started Free to install. MIT licensed. Yours to explore.