Cut your agent loose.

Open models by the second, one OpenAI-compatible API. Hold the token, get daily compute. No datacenter required.

get a keydocs →

$ curl run-space.com/v1/chat/completions -d '{"model":"llama-3.3-70b"}'

how it works

  1. 1

    Hold the token

    Your wallet's balance unlocks a daily compute allowance.

  2. 2

    Grab a key

    Connect your wallet, create an API key in the dashboard.

  3. 3

    Call a model

    One OpenAI-compatible request. Allowance resets daily.

Models

OpenAI-compatible endpoints for open models — Llama, Qwen, DeepSeek, and more. Point your existing client at one base URL.

curl https://run-space.com/v1/chat/completions \
  -H "Authorization: Bearer $RUNSPACE_KEY" \
  -d '{"model":"llama-3.3-70b","messages":[{"role":"user","content":"hi"}]}'

Rigs

A rig is a machine you connect. Point Claude Code at RunSpace and it allocates work — your local model handles the grunt work, the cloud takes the heavy reasoning.

# point Claude Code at RunSpace
export RUNSPACE_KEY=rsk_live_...
ccr code   # routes: local model for grunt work, cloud for the heavy lifting

Open models, live

7 available

One base URL, every model. Swap by changing a string.

Llama 3.3 70B Instruct
llama-3.3-70b
70B · 131k
Qwen3 235B-A22B
qwen3-235b
235B (MoE) · 262k
DeepSeek V4 Pro
deepseek-v4
MoE · 512k
Kimi K2.6
kimi-k2
MoE · 262k
GLM-5
glm-5
MoE · 202k
gpt-oss 120B
gpt-oss-120b
120B (MoE) · 131k
gpt-oss 20B
gpt-oss-20b
20B · 131k

Drop it into what you already use

It speaks the OpenAI API. Keep your client, your framework, your agent loop. Change two lines and you're on open models.

works with the OpenAI SDK, LangChain, Vercel AI SDK, anything OpenAI-shaped.
python
from openai import OpenAI

client = OpenAI(
    base_url="https://run-space.com/v1",
    api_key="$RUNSPACE_KEY",
)

resp = client.chat.completions.create(
    model="llama-3.3-70b",
    messages=[{"role": "user", "content": "ship it"}],
)

Pricing

No subscription. Hold the token and get a daily compute allowance that resets every day. Usage is metered per token, wholesale-plus.

50,000+ tokens
$1.00 / day compute
500,000+ tokens
$10.00 / day compute
5,000,000+ tokens
$50.00 / day compute

Tiers are placeholders until the token launches. No minimums, no lock-in.

Questions

Is it really OpenAI-compatible?
Yes. Point any OpenAI client at the base URL with your key. Same request and response shape, streaming included.
How does billing work?
Token-gated. Hold the token in your connected wallet and you get a daily compute allowance by tier. It resets every day. No subscription, no top-ups.
Which models?
Open weights only — Llama, Qwen, DeepSeek, Kimi, GLM, gpt-oss. The live list and ids are on the models page.
What are rigs?
A machine you connect. Point Claude Code at RunSpace and it routes work between your local model (Ollama) and the cloud — grunt work stays local and private, heavy reasoning goes to open models in the cloud.