Chat Completions
POST /v1/chat/completionsMessages in, message out.
Yes
Model
DeepSeek-V4.1-Flash-Abliterated
The DeepSeek-V4.1-Flash-1017 weights, edited to remove refusal behavior. Served over an OpenAI-compatible API on a B300 fleet.
For immediate use or interest, email info@belweave.com.
At a glance
| Label | Value |
|---|---|
| Model ID | deepseek-ai/DeepSeek-V4.1-Flash-1017 |
| Base model | DeepSeek-V4.1-Flash-1017 |
| License | MIT |
| Context length | 262K tokens |
| Modality | Text |
| API | OpenAI-compatible /v1/chat/completions |
| Base URL | https://api.infer.belweave.ai/v1 |
| Streaming | Server-sent events (SSE) |
| Hardware | B300 fleet |
| Availability | See Status. |
Method
What abliteration changes.
Abliteration removes refusal behavior from the model. belweave doesn't add an output filter.
Supported features
Streaming
stream: trueReceive tokens as server-sent events.
Tool calling
Get tool calls for your application to handle.
Reasoning output
Receive reasoning alongside the answer.
Use
Built for software and engineering.
- Building, reviewing, and refactoring software
- Debugging applications and infrastructure
- Designing systems, tests, and engineering tools
- Exploit development and malware analysis for legitimate purposes
- Authorized pentesting and bug-bounty research
- Detection rules and hardening guides