Dwight 4B 1.0
Dwight 4B 1.0 is a 4-billion-parameter model for analyzing sales-call transcripts, based on Qwen3-4B-Instruct-2507. It produces one-token decisions with calibrated confidence or structured JSON insights with timestamps and verbatim quotes.
What it does
Use it to classify sales calls with specific questions or generate structured, quotable call insights.
- Availability
- First release, version 1.0; model hosted on Hugging Face
- Pricing model
- Self-hosted inference costs are estimated in the announcement; no model price is stated
- Deployment
- Self-hosted on one 24–80 GB GPU; best with vLLM
- Privacy
- Not established
- Licence
- Apache 2.0
Capability
Modality
Access
Deployment
Pricing
Target user
Evidence
6 claims- Dwight 4B reads sales-call transcripts and answers typed questions with one token and a calibrated confidence. [1]
- It can produce structured JSON insights with a heading, detail, grade, score, timestamp, and verbatim excerpt. [1]
| Insights | JSON with heading, detail, grade, score, timestamp and a verbatim excerpt |
Hugging Face. Quoted text is outside CC BY 4.0 and keeps its original rights. - Its context length is 32,768 tokens. [1]
| Context | 32,768 tokens |
Hugging Face. Quoted text is outside CC BY 4.0 and keeps its original rights. - The announcement identifies version 1.0 as the first release and provides the model on Hugging Face. [1]
- The model runs on one 24–80 GB GPU and is best used with vLLM. [1]
| Hardware | one 24-80 GB GPU (bf16, ~8 GB of weights); best with vLLM |
Hugging Face. Quoted text is outside CC BY 4.0 and keeps its original rights. - The model is English-only and was trained on B2B software sales calls. [1]
- English only; trained on B2B software sales calls.
Hugging Face. Quoted text is outside CC BY 4.0 and keeps its original rights.
Sources
1
1
Hugging FaceOwner-controlled
https://huggingface.co/zime-ai/dwight-4b
Retrieved 2026-10-29 00:08 UTC
Verified
Versions
2| Version | Effective | Change | Changed fields | State |
|---|---|---|---|---|
| 2 | 2026-10-15 15:25 UTC | material change | capability, availability, pricing model, deployment | Current |
| 1 | 2026-10-14 16:22 UTC | publication | None | Current |
In briefs
1- Extract structured text from PDFs and images on a CPU Daily brief 2026-10-14
Original catalog content is licensed under CC BY 4.0, attributed to wowsuchai. Source quotations and linked content keep their original rights.