PodcastIntelligence Snacks · Weekly conversations about AI, software and business, from the team behind Normie Mode.Listen →
Models · Google DeepMind

Gemini 3.1 Pro

Google DeepMind's Gemini model.

Top three for

1 task, including documents, pdfs and notes.

Evidence

Independently tested. 34 independent results so far.

Cost

$12 per 1,000 typical requests, if you use it directly.

Reads at once

1,573 pages of about 500 words.

Where you can use it

The apps and plans that include it, cheapest first.

App and planPriceAccess
Gemini FreeFreeLimited
Limited use
Google AI Plus$4.99/monthLimited
Twice the free plan's use
Google AI Pro$19.99/monthIncluded
Google AI Ultra$99.99/month
From $99.99 a month; a $199.99 tier has higher limits
Included

Where it ranks, task by task

Its place on each task's score, among the models we compare that have enough results to score.

TaskPositionTask score
Best AI for documents, PDFs and notes3rd of 1373
Best AI for translation and languages4th of 1473
Best AI for writing4th of 1457
Best AI for planning and personal admin5th of 1470
Best AI for health questions5th of 1472
Best AI for studying and homework6th of 1274
Best AI for CVs, job applications and interviews6th of 1453
Best AI for maths and science7th of 1275
Best AI for research7th of 1152
Best AI for presentations and PowerPoint7th of 835
Best AI for business8th of 828
Best AI for legal questions8th of 838
Best AI for accounting and finance8th of 828
Best AI for Excel and spreadsheets8th of 849
Best AI for coding and building apps9th of 1016

Test results

Every test we use that has a result for this exact version.

TestResultPosition
Answering short factual questions correctly
Google DeepMind
74%2nd of 11
People's votes on health questions
LMArena
rating 14873rd of 13
Pulling information out of documents into tables
DTBench authors
97%3rd of 13
Graduate-level science questions
NYU and others
94%3rd of 11
People's votes on creative writing
LMArena
rating 14824th of 14
People's votes on questions in other languages
LMArena
rating 14754th of 14
People's votes overall
LMArena
rating 14804th of 14
People's votes on writing and language
LMArena
rating 14804th of 14
People's votes on back-and-forth conversations
LMArena
rating 14865th of 14
People's votes on legal questions
LMArena
rating 14885th of 14
People's votes on long, detailed requests
LMArena
rating 14825th of 14
People's votes on science questions
LMArena
rating 14985th of 14
People's votes on maths questions
LMArena
rating 14856th of 13
People's votes on business and finance questions
LMArena
rating 14597th of 14
Getting jobs done on a computer's command line
Terminal-Bench team
80%1st of 2
People's votes on following instructions
LMArena
rating 14658th of 14
Competition maths
Epoch AI
96%8th of 12
People's votes on coding questions
LMArena
rating 148310th of 14
People's votes on expert questions
LMArena
rating 148410th of 14
Humanity's Last Exam
Center for AI Safety, Scale AI
46%3rd of 4
Research-level maths
Epoch AI
60%9th of 11
Professional tasks in banking, consulting and law
Mercor
35%8th of 8
Building features in real software projects
Datacurve
12%8th of 8

Details

Maker
Google DeepMind
Released
Date not stated
Status
Current
Name checked
Confirmed with Google
Intelligence Snacks newsletter

The big AI ideas each week, in your inbox