abliteration.ai
Abliteration AI, Inc., Palo Alto: refusal-removed derivatives of open-weight models.
Abliteration AI builds abliterated variants of open-weight models: the base weights are modified and fine-tuned so the model no longer refuses cyber, red-team and agent-testing requests. Its current line pairs a general model with vision input and a large reasoning model built on GLM-5.3.
abliteration.aiCall abliteration.ai models through Opper with one API key. Sign up without a credit card.
Models from abliteration.ai
| Model | Context | Provider | Region | Zero data retention | Training | Input | Output | Cache read |
|---|---|---|---|---|---|---|---|---|
| 1M | US | No | $3.00 | $5.00 | $0.30 | |||
| 262K | US | No | $1.00 | $3.00 | $0.10 |
