Abliterated weights, served on our own hardware. It completes the request you made instead of explaining why it would rather not. Drop-in for the Anthropic and OpenAI APIs — change one environment variable and your tooling is already working with us.
Get an API key Compare the modelsThe workhorse. Fast, cheap, and it carries long agent chains without coming apart.
$4 / month · 1M context
Precise. It reasons more slowly and improvises less, but it is also the one that rarely breaks the code on step three.
$5 / month · 1M context
The longest and the most talkative. Holds a whole repository in context and stays coherent across long answers.
$6 / month · 1M context
Sign in with Google, copy the key, set the base URL. If you have used any third-party model endpoint before, this is the same thing.
// ~/.claude/settings.json — works on Windows, macOS and Linux
{
"env": {
"ANTHROPIC_BASE_URL": "https://abliterateme.com",
"ANTHROPIC_AUTH_TOKEN": "sk-your-key-from-the-dashboard",
"ANTHROPIC_MODEL": "glm-5.3"
}
}
# PowerShell $env:ANTHROPIC_BASE_URL = "https://abliterateme.com" $env:ANTHROPIC_AUTH_TOKEN = "sk-your-key-from-the-dashboard" $env:ANTHROPIC_MODEL = "glm-5.3" claude "refactor the parser and add tests"
Docs also have the bash form and the OpenAI-compatible client form.
| Trial | Standard | |
|---|---|---|
| Price | $0 — 14 days | $4–6 / month, see pricing |
| Models | all three | all three |
| Usage limits | 10× Claude Pro plan usage | 10× Claude Pro plan usage |
| Card required | No | Yes |
Yes, four dollars a month. We run our own hardware, the weights are
not new, and there is no frontier lab to pay for. The trade is capacity: the pool is
shared, and at peak hours trial keys wait behind each other — that is what the
429 on the status page means.