Technical report
Needle-1.0: a model for typed decisions
How a 395M-parameter model answers typed questions in one forward pass, what the experiments show, and where it struggles.
Current work · needle-1.0
Needle
A decision model that answers typed questions about a supplied state. Choices, scores, and probabilities in a single forward pass.
In development. The inference API is not yet available.
An example
Supply a situation and the questions you want answered. Each question defines its possible answers.
These are fixed examples showing the intended interaction, not live model inference.
1. Supplied state
Forum post: Selling 2 concert tickets, DM me. Payment by gift card only, no refunds!!
- action
- choice
- scam
- probability of true
- toxicity
- ordered score
Fixed example. No request is sent to a model.
2. Show illustrative response2. Illustrative response
All answers returned together in one forward pass.
actionchoice → flag
approve0.08
flag0.63
remove0.29
scamprobability of true → 0.86
P(true)0.86
toxicityordered score → 0.05
none0.95
mild0.05
severe0.00
3. Application policy
Here, flag is the selected action at probability 0.63, and the scam probability is 0.86. The application can send the post to a human reviewer rather than removing it automatically.
Inspect the proposed request
{
"model": "needle-1.0",
"state": "Forum post: Selling 2 concert tickets, DM me. Payment by gift card only, no refunds!!",
"questions": {
"action": {"type": "choice",
"criteria": {"approve": null, "flag": null,
"remove": null}},
"scam": {"type": "noul",
"instructions": "Is this likely a scam?"},
"toxicity": {"type": "score",
"criteria": ["none", "mild", "severe"]}
}
}The model
Needle reads the supplied state and scores the candidate answers directly. It does not generate text or search through a sequence of responses.
- Model
- needle-1.0
- Architecture
- Bidirectional encoder, ModernBERT-large backbone
- Parameters
- 395 million
- Answers
- Choice, Score, and Noul (probability of true)
- Computation
- One forward pass, zero output tokens
Evaluation
Needle performs well on the tested decision families. Unfamiliar task families and reordered choices remain weaknesses.
| Evaluation panel | Needle | Reference |
|---|---|---|
| Blind test1,180 fields | 97.0% | 85.8% |
| New wording and layouts432 fields | 98.1% | 88.2% |
| Unseen task families675 fields | 80.4% | 96.1% |
| Reordered choices1,008 fields | 82.9% | 93.9% |
Field accuracy on four named panels, using the same requests for both models. One checkpoint and one seed, not a general reasoning score. The checkpoint label is a readable alias, not an artifact hash.
Research
Follow the work
Occasional updates on Needle’s development and new results.