Decision model

Drex Max 1.1

1 question per request

67.4% of the 2,198 questions right (65.4% to 69.3%, 95% interval). Back to the leaderboard.

How we called it

Kind: Decision model
Access: Nace.AI System One API
Paid as: Provider service
Model ID: drex-max-v1.1
Note: Nace.AI's hosted Drex Max v1.1 (drex-max-v1.1, released 28 October 2026), a newer checkpoint than the open Drex Max v1 weights; Nace publishes no weights for it. Nace bills $0.03 per million input tokens during its launch, half the $0.06 list price the cost here uses.
Confidence from: Option score
Run: 326, 31 October 2026

Settings

Reasoning: off
Request format: 1 question per request
Requests at once: 1
Temperature: not recorded
Harness seed: 0
Harness commit: 61f3824138e9
Prompt version: 254597b8be16

Results

Questions	Accuracy	95% interval	Right
All questions	67.4%	65.4% to 69.3%	1,481 of 2,198
Reading	88.6%	85.9% to 90.8%	581 of 656
Grammar	58.4%	55.9% to 60.8%	900 of 1,542

Comparison group: pqr. Compare with another setup.

Accuracy by topic

Topic	Part	Accuracy	Right
Word and phrase meanings	Reading	84.7%	111 of 131
Sentence meanings	Reading	93.5%	144 of 154
How a writer explains and develops ideas	Reading	84.3%	59 of 70
Topic and main point of a paragraph	Reading	93.9%	92 of 98
How a paragraph fits together	Reading	81.1%	86 of 106
Details that support a paragraph’s main point	Reading	91.8%	89 of 97
Word types, such as nouns and verbs	Grammar	65.7%	92 of 140
Groups of words built around a noun	Grammar	56.9%	37 of 65
Verbs and their forms	Grammar	62.6%	57 of 91
Verb forms used like nouns or modifiers	Grammar	68.1%	49 of 72
Word endings and what they do	Grammar	64.6%	51 of 79
How words are built	Grammar	43.5%	30 of 69
The roles of words and phrases in a sentence	Grammar	53.8%	78 of 145
How verbs show who acts and who receives the action	Grammar	60.8%	45 of 74
Types of sentences	Grammar	65.9%	56 of 85
Questions combining several grammar topics	Grammar	46.4%	102 of 220
Sound patterns and changes	Grammar	33.8%	26 of 77
Spelling	Grammar	61.7%	82 of 133
Punctuation	Grammar	61.5%	72 of 117
Problems in sentence wording	Grammar	70.3%	123 of 175

Page images (preliminary)

Preliminary: these numbers are not final and may change. This setup is condition A of the Page images tab, which asks the 763 questions with underlined or numbered words as text (A), as a page image (B), as both (C) and through OCR models (D); the share answered right, with its 95% interval.

Model	A · text onlyaccuracy, higher is better	B · image onlyaccuracy, higher is better	C · text plus imageaccuracy, higher is better	Best OCR routeaccuracy, higher is better
Drex Max 1.11 question per request	64.1%60.6% to 67.4%	45.1%41.6% to 48.6%	65.5%62.1% to 68.8%	Not available: not run

Confidence

Answers kept at 99% or more: 220, of which 0 wrong (share right: 98.3% to 100.0%, 95% interval); 1,978 skipped, the model less than 99% sure
Calibration gap: 2.9 points
AUROC, all questions: 0.83 0.82 to 0.85
AUROC, reading: 0.92 0.90 to 0.95
AUROC, grammar: 0.77 0.75 to 0.79
Brier score: 0.154
Accuracy of its surest answers: most confident fifth: 99.5% right; most confident half: 90.2% right; most confident four fifths: 76.0% right
Typical score of a right answer: 84.9%
Typical score of a wrong answer: 46.6%

Repeats

Run 5 times on all questions.

Answers changed in 5 runs: 0 of 2,198
Same wrong answer in all 5: 717 of 2,198
Wrong in at least one run: 717 of 2,198
Accuracy in each run: 67.4%, 67.4%, 67.4%, 67.4%, 67.4%

Latency

Latency, median request: 0.313 s (this run)
Latency, 90th percentile: 0.449 s
Requests at once: 1
Throughput: 119.0 questions per minute of request time
Request time: 18.5 minutes, over the requests whose answers were kept

Latency: the time from sending one request to receiving its answer, measured from our computer, network and any retries within the request included. Request time: the time at least one request was in flight, over the requests whose answers were kept. It is not the run's elapsed time: it leaves out requests that failed and any pause before a run was resumed.

Cost

$ per 1,000 questions: $0.023
Whole run: $0.051
Basis: Estimated from published prices and recorded token usage
Price: $0.06 input, output free per 1M tokens (source, 2026-10-30)
Input tokens per question: 385
Output tokens per question: 163

Going first in front of an LLM

Replay of recorded answers, cutoff chosen on half the questions, averaged over the splits. Savings are estimates from API prices; the Method page explains how.

LLM behind it	Share taken	Accuracy change	Saving
Gemini 3.8 Flashhigh reasoning · 16 questions per request	18.2%	−0.10 points	17%
GPT-6 Astrahigh reasoning · 16 questions per request	20.8%	−0.13 points	20%
GPT-6.1 Solhigh reasoning · 16 questions per request	23.5%	−0.19 points	22%
GPT-6 Astralow reasoning · 16 questions per request	21.8%	−0.16 points	21%
GPT-6 Astralow reasoning · 1 question per request	19.6%	−0.14 points	19%
Gemini 3.8 Flashlow reasoning · 1 question per request	21.5%	−0.13 points	19%
GPT-6.1 Sollow reasoning · 1 question per request	26.2%	−0.22 points	26%
GPT-6.1 Sollow reasoning · 16 questions per request	22.0%	−0.16 points	20%
Claude Opus 5.5high reasoning · 16 questions per request	23.2%	−0.19 points	23%
GPT-6.1 Terrahigh reasoning · 16 questions per request	22.0%	−0.17 points	18%
Claude Opus 5.5low reasoning · 1 question per request	24.3%	−0.18 points	24%
Gemini 3.8 Flashlow reasoning · 16 questions per request	19.3%	−0.13 points	11%
GPT-6.1 Terralow reasoning · 16 questions per request	29.7%	−0.17 points	24%
Claude Opus 5.5low reasoning · 16 questions per request	33.8%	−0.24 points	33%
Claude Sonnet 5.5high reasoning · 1 question per request	35.5%	−0.28 points	35%
Claude Sonnet 5.5high reasoning · 16 questions per request	50.4%	−0.12 points	50%