Chat model

Gemini 3.8 Flash

low reasoning · 1 question per request

97.1% of the 2,198 questions right (96.3% to 97.8%, 95% interval). Back to the leaderboard.

We also ran this setup at 16 questions per request: Gemini 3.8 Flash · low reasoning · 16 questions per request.

Which decision model can go first in front of it? See the Routing tab.

How we called it

Kind: Chat model
Access: Gemini API
Paid as: Provider service
Model ID: gemini-3.8-flash
Confidence from: Written estimate
Run: 188, 13 October 2026

Settings

Reasoning: low
Request format: 1 question per request
Requests at once: 1
Temperature: not recorded
Harness seed: 0
Harness commit: 54043fd92cd5
Prompt version: 254597b8be16

Thinking levels (Google's thinking documentation, read 28 October 2026): low, medium (default), high. Top level: high. This run: low.

Results

Questions	Accuracy	95% interval	Right
All questions	97.1%	96.3% to 97.8%	2,135 of 2,198
Reading	96.2%	94.4% to 97.4%	631 of 656
Grammar	97.5%	96.6% to 98.2%	1,504 of 1,542

Comparison group: abcd. Compare with another setup.

Accuracy by topic

Topic	Part	Accuracy	Right
Word and phrase meanings	Reading	96.2%	126 of 131
Sentence meanings	Reading	98.7%	152 of 154
How a writer explains and develops ideas	Reading	91.4%	64 of 70
Topic and main point of a paragraph	Reading	98.0%	96 of 98
How a paragraph fits together	Reading	92.5%	98 of 106
Details that support a paragraph’s main point	Reading	97.9%	95 of 97
Word types, such as nouns and verbs	Grammar	97.9%	137 of 140
Groups of words built around a noun	Grammar	98.5%	64 of 65
Verbs and their forms	Grammar	100.0%	91 of 91
Verb forms used like nouns or modifiers	Grammar	98.6%	71 of 72
Word endings and what they do	Grammar	96.2%	76 of 79
How words are built	Grammar	95.7%	66 of 69
The roles of words and phrases in a sentence	Grammar	97.9%	142 of 145
How verbs show who acts and who receives the action	Grammar	100.0%	74 of 74
Types of sentences	Grammar	97.6%	83 of 85
Questions combining several grammar topics	Grammar	97.7%	215 of 220
Sound patterns and changes	Grammar	96.1%	74 of 77
Spelling	Grammar	97.7%	130 of 133
Punctuation	Grammar	95.7%	112 of 117
Problems in sentence wording	Grammar	96.6%	169 of 175

Confidence

Answers kept at 99% or more: 1,365, of which 6 wrong (share right: 99.0% to 99.8%, 95% interval); 833 skipped, the model less than 99% sure
Calibration gap: 0.6 points
AUROC, all questions: 0.84 0.79 to 0.89
AUROC, reading: 0.89 0.84 to 0.94
AUROC, grammar: 0.81 0.74 to 0.87
Brier score: 0.024
Accuracy of its surest answers: most confident fifth: 99.6% right; most confident half: 99.6% right; most confident four fifths: 98.8% right
Typical score of a right answer: 100.0%
Typical score of a wrong answer: 95.0%

Repeats

We did not repeat this setup.

Latency

Latency, median request: 2.72 s (this run)
Latency, 90th percentile: 4.67 s
Requests at once: 1
Throughput: 19.1 questions per minute of request time
Request time: 114.9 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: $0.96
Whole run: $2.12
Basis: Estimated from published prices and recorded token usage
Price: $0.75 input, $0.075 cached input, $3.75 output per 1M tokens (source, 2026-10-08)
Input tokens per question: 310
Output tokens per question: 195