Nace.AI Drex v1.5
Nace.AI Drex v1.5 by Nace.AI is a decision model: it answers typed questions with a probability for each allowed answer instead of generating text. It scores 49.7 on the Jev Decision Index (rank 46). Hosted, it costs about $0.010 per 1,000 decisions.
Pricing
| Provider | Input / 1M | Output / 1M |
|---|---|---|
| nace | $0.05 | $0.00 |
| openrouter | $0.04 | $0.00 |
Cost per 1,000 decisions assumes 200 input tokens per decision.
Specifications
- Provider
- Nace.AI
- Parameters
- 9B
- Released
- Sep 2026
- Open weights
- Yes
- Question types
- Choice (one of N options), Score (ordered rubric), Yes / no
- Mechanism
- Decoder LM with a decision readout
- Backbone
- XiaomiMiMo/MiMo-V2.6-Distill-Qwen-9B
- Jev-compatible request
- Yes
- Runs on CPU
- No
- License
- nace-ai-open-rail-m
- Hugging Face
- nace-ai/drex-v1.5
Benchmarks
- Jev Decision IndexECE 0.090
- 49.7
Jev Decision Index: public 58.1, ECE 0.090, Brier 0.383. Lower ECE means its probabilities can be taken at face value.
Self-reported by the publisher
The model's own claims, often on its own sets or a self-run copy of a public suite. Not independently verified and never ranked alongside measured results.
- Jev Decision Index public suite (self-run)Self-reportedDecision Index public balanced skill, edition 0.3 kit (release-v3), self-run
- 58.1
Speed & latency
Compare Nace.AI Drex v1.5 with…
FAQ
How much does Nace.AI Drex v1.5 cost?
Hosted, Nace.AI Drex v1.5 costs $0.05 per 1M input tokens, about $0.010 per 1,000 decisions at 200 input tokens each.
What questions can Nace.AI Drex v1.5 answer?
Nace.AI Drex v1.5 answers choice (one of n options), score (ordered rubric), yes / no questions, returning a probability for each allowed answer.
Is Nace.AI Drex v1.5 well calibrated?
On the Jev Decision Index its expected calibration error (ECE) is 0.090; below about 0.05, a stated 80% confidence is right roughly 80% of the time.
Is Nace.AI Drex v1.5 open weight, and can it run on a CPU?
Yes, Nace.AI Drex v1.5 has open weights. It needs a GPU to run at useful speed.
Data as of 2026-10-14. Source: Metatext decision-model catalogue, 2026-10-14 (OpenRouter, Jev Decision Index, Hugging Face). Pricing is per 1M tokens (USD); confirm with the provider before production use.