Connecting to cluster…

GPU Inference Factory

Self-hosted 3× H100 cluster (Jetstream2 ACCESS allocation, through 2026-09-25) serving frontier open models. OpenAI-compatible API. Zero cost during beta.

Start Building — Free View Docs
–
H100 Online
–
Tokens Served
131K
Context Window
100%
Test Pass Rate (2026-09-21)
$0
Beta Cost

GPU Infrastructure

GPU pool status from our 3× H100 cluster (ACCESS allocation through 2026-09-25) · updated …

Why Token Factory

Everything you need to build and ship LLM-powered applications

⚡

Self-Hosted Infrastructure

Our own H100 cluster (ACCESS allocation through 2026-09-25). No cloud markup, no rate limits from third parties. Full control over serving, caching, and routing.

🔄

Drop-in OpenAI Replacement

Point your existing OpenAI client at our base_url. No code changes required. Works with Claude Code, Cursor, and any OpenAI-compatible tool.

🧠

Frontier Open Models

Qwen 3.8-27B for speed and quality. GPT-OSS 120B for deep reasoning. Uncensored for research flexibility.

📊

GPU Status Dashboard

Fleet status straight from live probes. See GPU utilization, VRAM, latency, and concurrent sessions — rendered honestly when everything is down. Full transparency.

🛠️

Tool Calling

Native function calling with precise tool selection. Multi-tool orchestration verified at 100% accuracy.

💰

Free During Beta

No credit card. No cost. We burned our Jetstream2 ACCESS allocation (through 2026-09-25) to give the research community free inference.

Available Models

Frontier open models behind a single unified API

ModelContextStrengthStatus
qwen-3.8-27b 131K Coding, tools, fast inference ● Live (as of 2026-09-21)
gpt-oss-120b 131K Deep reasoning, analysis ● Live (as of 2026-09-21)
qwen-72b 131K Higher quality coding ◐ Coming Soon

Get Started in Three Steps

From zero to inference in under 2 minutes

01

Sign Up

Create a free account with your email. No credit card needed.

02

Get Your API Key

Generate a key from your dashboard in one click.

03

Call the API

Use any OpenAI-compatible client. Just change the base URL.

One curl Away

Use the same client libraries you already know

# Works with any OpenAI-compatible client
curl https://api.agent-next.com/v1/chat/completions \
  -H "Authorization: Bearer $TOKEN_FACTORY_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "qwen-3.8-27b",
    "messages": [{"role": "user", "content": "Hello!"}]
  }'