---
title: "Artificial Analysis Decision Index: how Drex Max ranks"
description: "Drex Max v1.1 scores 72.6 on the Artificial Analysis Decision Index v2.1, the highest of 18 decision models. What the index measures, how it is scored, and where Drex leads."
canonical: "https://www.nace.ai/drex/decision-index"
updated: 2026-11-12
---

# Artificial Analysis Decision Index: how Drex Max ranks

> Drex Max v1.1 scores 72.6 on the Artificial Analysis Decision Index v2.1, the highest of 18 decision models. What the index measures, how it is scored, and where Drex leads.

The Artificial Analysis Decision Index is an independent benchmark of decision models: models that answer by choosing from a fixed set of options, with a probability for each. Artificial Analysis runs every model itself on six evaluations and combines them into one score. On v2.1, Drex Max v1.1 by Nace.AI scores 72.6, ahead of GLiDE v1.2 at 72.5 and every other decision model.

Drex Max v1.1: 72.6. The highest score of the 18 decision models on the index. GLiDE v1.2 is next at 72.5, at $780 per million tasks against $103 for Drex Max v1.1.

Source: [Artificial Analysis Decision Index v2.1, as of Nov 12, 2026](https://artificialanalysis.ai/evaluations/artificial-analysis-decision-index).

## Where the 72.6 comes from

No decision model wins every evaluation. Drex Max v1.1 ranks #1 of 18 decision models on AA-Paperclip, the most heavily weighted part of the index at 30%, and no lower than #5 on the other 5. That balance puts it first among decision models.

| Evaluation | Weight | Drex Max v1.1 | Rank of 18 | Best decision model |
| --- | ---: | ---: | ---: | ---: |
| AA-Paperclip | 30% | 72.8 | #1 | 72.8 |
| EvalSafe-O*NET-AA | 15% | 88.3 | #5 | 91.1 |
| AA-Omniscience-Decision | 15% | 60.3 | #5 | 66.3 |
| HLE-MultiChoice | 15% | 46.6 | #5 | 48.7 |
| AA-Symmetry | 15% | 79.8 | #4 | 88.2 |
| When2Call | 10% | 95.3 | #3 | 95.8 |

Rank and best score are among the 18 decision models.

## The top decision models

| Rank | Model | Creator | Decision Index |
| ---: | --- | --- | ---: |
| 1 | Drex Max v1.1 | Nace.AI | 72.6 |
| 2 | GLiDE v1.2 | Fastino | 72.5 |
| 3 | Mammoth 27B | Base Labs | 71.4 |
| 4 | d1-ultra | Liquid AI | 70.6 |
| 5 | Decider v1.2 | Perplexity | 69.8 |

Ranked among the 18 decision models on the index. Artificial Analysis can also plot language models alongside them for reference.

[Full leaderboard on Artificial Analysis](https://artificialanalysis.ai/evaluations/artificial-analysis-decision-index)

## What the Decision Index measures

Each answer gets a Decision Score of 50 × (1 + c − (p − c)²), where p is the probability the model gave its chosen answer and c is 1 if that answer is right. A right answer scores between 50 and 100, a wrong one between 50 and 0, so a model loses most when it is confidently wrong. Artificial Analysis describes it as the payoff of acting on answers above a confidence threshold and routing the rest, averaged over every threshold.

- AA-Paperclip (30%): decisions and extraction over resumes, receipts, invoices and research papers
- EvalSafe-O*NET-AA (15%): judgement about workplace documents, from TypeSafe's O*NET set
- AA-Omniscience-Decision (15%): knowledge, and declining instead of guessing
- HLE-MultiChoice (15%): reasoning on multiple-choice questions from Humanity's Last Exam
- AA-Symmetry (15%): whether a decision holds when options are reordered or reworded, or the question is negated
- When2Call (10%): whether to call a tool, and which, from NVIDIA's evaluation

AA-Symmetry is scored differently: a model gets nothing if its top option changes when the question is reworded, and otherwise loses the range of its confidence across the versions.

Cost is the average cost per task at current prices, times one million. Decision models are billed for input tokens only; language models for input and output. Time is the median time per task through each provider's API, in milliseconds. Both are averaged across the evaluations with the same weights.

## What is a decision model?

A decision model reads a case and a fixed set of options and returns a probability for each option in one pass. A language model writes text that has to be parsed into an answer. Because a decision model returns probabilities instead of tokens, it is fast and cheap to run, it cannot answer outside the options, and how sure it is can be measured. That is what the Decision Index scores.

## Drex Max vs Jev 1.14

Jev 1.14 is TypeSafe's decision model. On the Artificial Analysis Decision Index v2.1, as of Nov 12, 2026, Drex Max v1.1 scores 72.6 against 66.2. Jev 1.14 is ahead on EvalSafe-O*NET-AA and answers faster, at 92 ms per task against 283 ms.

|  | Drex Max v1.1 | Jev 1.14 |
| --- | ---: | ---: |
| Decision Index | 72.6 | 66.2 |
| AA-Paperclip | 72.8 | 57.5 |
| EvalSafe-O*NET-AA | 88.3 | 91.1 |
| AA-Omniscience-Decision | 60.3 | 60.1 |
| HLE-MultiChoice | 46.6 | 44.5 |
| AA-Symmetry | 79.8 | 74.5 |
| When2Call | 95.3 | 84.4 |

[Switch from Jev to Drex](https://www.nace.ai/migrate-to-drex)

## The Public Decision Index (0.4.1)

The Public Decision Index is a different leaderboard: a community board on Hugging Face, named after Jev, that scores chance-corrected decisions across public benchmark suites. It is versioned by edition rather than by date. On edition 0.4.1, where Drex Max is listed as Drex Max v1, its public score is 67.5 against 60.7 for Jev 1.14.

- [Drex's full Public Decision Index results](https://www.nace.ai/drex#decision-index)
- [Public Decision Index on Hugging Face](https://huggingface.co/spaces/multimodalart/jev-decision-index)

## Frequently asked questions

### What is the Decision Index?

The Decision Index is Artificial Analysis' benchmark for decision models. It combines six evaluations into one score out of 100: AA-Paperclip (30%), EvalSafe-O*NET-AA (15%), AA-Omniscience-Decision (15%), HLE-MultiChoice (15%), AA-Symmetry (15%), When2Call (10%). Version v2.1 lists 18 decision models and 4 language models for reference.

### What is a decision model?

A decision model is an AI model that answers by choosing from a fixed set of options and returns a probability for each, in one call. A language model writes text that is parsed into an answer. Decision models are faster and cheaper per task, cannot answer outside the options, and their confidence can be measured, which is what lets you act on sure answers and route the rest to a person.

### How is the Artificial Analysis Decision Index scored?

Each answer is scored 50 × (1 + c − (p − c)²), where p is the probability given to the chosen answer and c is 1 when it is right. Confident mistakes cost the most. The index is a weighted average of that score on five evaluations and a consistency score on AA-Symmetry. Cost per million tasks and median time per task are reported alongside it.

### Which decision model ranks highest on the Decision Index?

As of Nov 12, 2026 (v2.1), the top decision models are Drex Max v1.1 (72.6), GLiDE v1.2 (72.5) and Mammoth 27B (71.4). Drex Max v1.1 costs $103 per million tasks against $780 for GLiDE v1.2.

### How does Drex compare to Jev on the Decision Index?

Drex Max v1.1 scores 72.6 against 66.2 for Jev 1.14 and leads it on AA-Paperclip, AA-Omniscience-Decision, HLE-MultiChoice, AA-Symmetry and When2Call. Jev 1.14 leads on EvalSafe-O*NET-AA and is faster per task. Drex's API accepts Jev-style requests, so switching takes three settings.

### Is it the same as the Public Decision Index 0.4.1?

No. The Public Decision Index is a separate community leaderboard on Hugging Face, versioned by edition (the current one is 0.4.1). The Artificial Analysis Decision Index is run independently by Artificial Analysis and versioned v2.1. Drex Max is ahead of Jev 1.14 on both.

### How do I try Drex Max?

Drex Max is available through the Nace Console API under the model id drex-max, and on OpenRouter. Create an API key in the Console and send a case with its options; each answer comes back as a probability per option.

[Drex](https://www.nace.ai/drex)

## For agents

- [Drex API guide](https://www.nace.ai/drex.md): the decision model: request shape, question types, benchmarks and deployment, as markdown
- [Decision endpoint](https://console.nace.ai/v1/systemone): POST with model "drex-latest", a state string and typed questions (choice, noul, score); https://console.nace.ai/v1/models lists checkpoints
- [Document endpoints](https://console.nace.ai/v1/documents/{parse|split|classify|extract|ground}): POST a source (a public URL, an uploaded workspace file or a finished parse job) and poll GET https://console.nace.ai/v1/documents/jobs/{id}; the same key and credit as decisions
- [Console](https://console.nace.ai): API keys, the playground, usage and billing; send `Authorization: Bearer $NACE_API_KEY` on every request
- [Console docs](https://console.nace.ai/docs): quickstart, question types, the document routes, errors, limits and pricing
- [Nace Document Processing](https://www.nace.ai/perception.md): the Perception series: parse, split, classify, extract and ground any file type with a location on every value
- [Drex for Search](https://www.nace.ai/drex/search.md): Drex for Search: decision-powered retrieval for agents, Drex finds the page, cited search through the CLI, SDKs, MCP and REST, and the public benchmark ranks
- [Industries](https://www.nace.ai/industries.md): Nace by industry: the files Document Processing reads and the decisions Drex makes for accounting, banking, mortgage, insurance, legal, healthcare, government, construction, logistics and professional services
- [Developers](https://www.nace.ai/developers.md): The execution layer for agents: Perception, Decision and Search behind one key, the Drex Max frontier, install steps, a scripted agent session and the four deployment boundaries
- [Site index](https://www.nace.ai/llms.txt): every page on nace.ai; append `.md` to any page URL or send `Accept: text/markdown` for the markdown version