DeepInfra raises $107M Series B to scale the inference cloud — read the announcement
nace-ai/
$0.06 / 1M tokens
Drex v2.0 is a 35B mixture-of-experts decision model (3B active) from Nace.AI. Given a state (text or JSON) and typed questions (yes/no, multiple choice, or ordinal score), it returns a calibrated probability for every option in a single forward pass, with no text generation. It is TypeSafe System One compatible, handles contexts up to 131K tokens, and ranks #1 on Decision Index 0.2.1 with 60.78, ahead of Drex v1.5 and Jev 1.13.0.

Log in to run decisions.
Drex v2.0 is a decision model from Nace.AI. You give it a state (text or JSON) and typed questions, and it returns a calibrated probability for every option in a single forward pass. It generates no text, and it never answers with an option you did not supply.
| Type | You send | You get back |
|---|---|---|
noul | a yes/no question, with optional criteria.true / criteria.false | noul, the probability of "yes" |
choice | options as criteria (name → description or null), 1-255 options | choice (the most likely option), probabilities per option, and confidence |
score | ordered levels as a criteria list, 1-255 levels | score (the expected level), probabilities per level, legend, and confidence |
One request can ask many questions about the same state. The state is read once and billed once per request.
curl https://api.deepinfra.com/v1/decisions \
-H "Authorization: Bearer $DEEPINFRA_TOKEN" \
-H "Content-Type: application/json" \
-d '{
"model": "nace-ai/drex-v2.0",
"state": {"ticket": "I was charged twice for order A-104. Please refund the duplicate today."},
"questions": {
"team": {"type": "choice", "instructions": "Which team handles this?",
"criteria": {"billing": "Charges and refunds", "shipping": "Delivery", "other": null}},
"refund": {"type": "noul", "instructions": "Does the customer ask for a refund?"},
"urgency": {"type": "score", "instructions": "How urgent is this request?",
"criteria": ["Routine", "Time-sensitive", "Immediate"]}
}
}'
The same API is served at https://api.deepinfra.com/v1/systemone. To use the TypeSafe Python SDK, set TYPESAFE_BASE_URL=https://api.deepinfra.com/typesafe and model="nace-ai/drex-v2.0".
Figures reported by Nace.AI on the model card.
| Rank | Model | Index | Knowledge & Reasoning | Language | Retrieval | Tools | Arts |
|---|---|---|---|---|---|---|---|
| 1 | Drex v2.0 | 60.78 | 47.7 | 65.9 | 62.9 | 76.0 | 49.1 |
| 2 | Drex v1.5 | 58.28 | 44.2 | 62.5 | 62.0 | 75.3 | 45.1 |
| 3 | Jev 1.13.0 | 57.91 | 51.4 | 62.0 | 55.4 | 75.1 | 37.7 |
Area columns are chance-corrected skill × 100. Drex v2.0 is ahead of Jev on 21 of the 38 index benchmarks and on 4 of 5 areas.
| Benchmark | Drex v2.0 | Drex v1.5 | Jev 1.13.0 |
|---|---|---|---|
| DecideBench v1.1 (400 contrastive decisions) | 96.3% | 90.0% | 98.0% (AI Space) |
| JevBench (231 public items) | 85.7% | 85.7% | 86.6% |
| Email classification, AUC | 0.984 | 0.972 | 0.981 |
| Email classification, caught at 5% false positives | 89.4% | 80.3% | 86.4% |
| Game arena vs Jev (256 games) | 60.9% score (132 W / 48 D / 76 L) | 56.8% | — |
Drex v2.0 is © Nace.AI, Inc. and is provided under the Nace AI Open RAIL-M License (modified). By using this model you agree to its use-based restrictions (Attachment A), including:
Read the full license before use. Drex and Nace are trademarks of Nace.AI, Inc.
© 2026 DeepInfra. All rights reserved.