Drop-in by design
Keep the OpenAI SDK, client, and request shape you already use. Change the base URL and key, then continue building.
Chat Completions · Responses · Streaming
Founder alpha now forming
Stop counting tokens.Start building.
Private, long-context Qwen3.8 27B through the API shape your tools already understand. Predictable access, honest limits, and no prompt or response retention in our gateway.
OpenAI compatibleStreamingTool callsLong context
import OpenAI from "openai";
const term0 = new OpenAI({
baseURL: "https://api.term0.dev/v1",
apiKey: process.env.TERM0_API_KEY
});
const stream = await term0.chat.completions.create({
model: "qwen3.8-27b",
messages, stream: true
});
A sharper contract
Power without the usual fine print.
Term0 is built around a small set of promises we can explain, measure, and keep.
Context stays yours
Conversation state lives in your application. The Term0 gateway forwards each request without creating a hidden conversation archive.
Client-side state · Long-context beta
Fair use, plainly stated
No per-token billing. Interactive use stays fast, while sustained power users are slowed when shared capacity is under pressure.
Predictable price · Transparent throttling
The engine
One model, selected on purpose.
Qwen3.8 27B gives Term0 a practical balance of code quality, reasoning, speed, and long-context capability without hiding a random model router behind one name.
See The API SurfaceTERM0 CORE / 01 active
Qwen3.827B
Context128K beta
PrecisionFP8
TransportSSE
ToolsNative
The full 128K window remains in staged validation during founder alpha.
Privacy, precisely
We meter traffic.
We do not warehouse thought.
Reliability needs counters. It does not need your source code, private questions, customer records, or generated answers.
Gateway data boundaryv1.0
Kept
Key ID, route, model, token counts, status, latency
Not kept
Read The Plain-English Privacy NotePrompts, responses, files, tool payloads, authorization headers
FOUNDING COHORT / LIMITED SEATS
Help shape the API you actually want to pay for.
The first cohort gets unmetered interactive access, direct input on limits and compatibility, and launch pricing locked before any charge.
- 01 No per-token billing
- 02 One active generation per account
- 03 Transparent pressure throttling
- 04 Direct founder support
Unmetered is not unlimited automation. Fair-use and abuse protections apply.
Request An Invitation01 / 100
Straight answers
No Mythology Required.
01Does unmetered mean literally unlimited?+
It means there is no per-token charge and no surprise token bill. Concurrency, rate, abuse, and shared-capacity protections still apply. Term0 is built for generous interactive use, not unlimited unattended automation.
02Do you store prompts or responses?+
The Term0 gateway does not store prompt text, response text, uploaded content, tool payloads, or authorization headers. It keeps minimal operational counters needed to protect the service and show usage.
03Will my existing OpenAI client work?+
If it lets you set a custom base URL, usually yes. The founder alpha supports OpenAI-style Chat Completions and Responses, streaming, and tool calls. Anthropic Messages compatibility is also available for selected clients.
04What happens when the service is busy?+
Every founder account gets a protected interactive lane. Heavy sustained usage may receive a clear 429 response or reduced throughput until pressure falls. We publish the rule instead of hiding it in vague unlimited language.