The jgram API
jgram, the self-improving AI model.
The jgram API is an OpenAI-compatible API for jgram, a frontier-level AI system powered by Bittensor.
Updated
Frontier quality for $0.40 per million input tokens and $1.95 per million output. OpenAI-compatible: change two lines and your code runs on jgram.
$5 of free credit when you sign up.
https://tao1.ai/v1lucyHow do I start using the jgram API?
Create a key in Account → API keys (it starts with jmk_), put it in TAO1_API_KEY, and send a message.
curl https://tao1.ai/v1/chat/completions \
-H "Authorization: Bearer $TAO1_API_KEY" \
-H "Content-Type: application/json" \
-d '{"model": "lucy", "messages": [{"role": "user", "content": "Hello, Lucy"}]}'
import os
from openai import OpenAI
client = OpenAI(base_url="https://tao1.ai/v1", api_key=os.environ["TAO1_API_KEY"])
print(client.chat.completions.create(model="lucy", messages=[{"role": "user", "content": "Hello, Lucy"}]).choices[0].message.content)
import OpenAI from "openai";
const client = new OpenAI({ baseURL: "https://tao1.ai/v1", apiKey: process.env.TAO1_API_KEY });
console.log((await client.chat.completions.create({ model: "lucy", messages: [{ role: "user", content: "Hello, Lucy" }] })).choices[0].message.content);
Python needs pip install openai; JavaScript needs npm install openai (run as an ES module). Streaming, tool calls and JSON mode work the same way as with OpenAI.
How much does the jgram API cost?
| Model | Input | Cached input | Output |
|---|---|---|---|
jgram (lucy) | $0.40 | $0.12 | $1.95 |
Prices are per million tokens and the same ones GET /v1/models returns. For comparison, Moonshot AI publishes Kimi K3 at $3.00 input, $0.30 cached input and $15.00 output per million tokens (platform.kimi.ai/docs/pricing/chat, checked October 14, 2026): jgram's input and output prices are each under one seventh of that.
How does jgram compare with the best models?
In our October 2026 runs, jgram placed #1 of 8 models on SimpleQA (jgram with web search, the other models without), 2nd of 8 on AA-Omniscience, and scored ahead of Kimi K3.1 on AutomationBench and TerminalBench 4.0 (jgram measured in our test harness). The full table, test conditions and raw results are on the jgram page. Earlier results (Oct 2026).
How does jgram work?
The right playbook for each request
For every request, jgram picks a playbook: the best open-weight model for the job, plus the harness, instructions and engines that go with it.
It checks its own work
Verifiers and repair passes run before the answer reaches you. Code is run and tested, and apps are opened in a real browser.
It improves every day
Real use updates the playbooks and tests daily, so jgram gets better without waiting for a new model release.
Which jgram model should I use?
| Model id | Use it when | Price (in / cached / out per M) |
|---|---|---|
lucy | Every request: balanced quality, cost and speed. | $0.40 / $0.12 / $1.95 |
- Streaming
- Tool calls
- JSON mode
- Images in
- PDF, Word, Excel and slide files
- 131K context
Does jgram work with my tools?
Anything that speaks the OpenAI API: set the base URL to https://tao1.ai/v1 and the model to lucy. The older id tao-1 is still accepted.
- OpenAI Python SDK
- OpenAI JavaScript SDK
- Kilo Code
- Cline
- Cursor
- Any OpenAI-compatible tool
Setup for Kilo Code and Cline is in the docs.
What happens to my data?
We do not keep the text of API prompts or answers, and we never use API requests to improve jgram. jgram sends your request to the AI provider that answers it. Providers handle data under their own policies, and some may keep it or use it to improve their models. Don’t share anything you wouldn’t want a third-party provider to see. Images you send go to the model for that request only and are not stored. The details are in our API privacy page.
Is jgram right for you?
A good fit if you already use an OpenAI-compatible client and want a lower price per token than Kimi K3.
Test it on your own tasks first. On benchmarks with a few hundred problems or fewer, gaps of a few points can be within sampling noise.
Common questions about the jgram API
How am I billed?
From the prepaid credits on your account. New accounts get $5 free; add more on the account page, where your balance and usage are listed. A request your balance can't cover answers 402 and is not sent to jgram.
What if a model or provider goes down?
jgram fails over automatically to other models and providers, so your request is still answered.
Can I choose the model jgram uses?
Yes. Most settings, including which model leads, can be set per request; see the docs. By default jgram picks for you.