Google · not tested yet
Gemini 3.1 Pro in Taskade
Gemini 3.1 Pro from Google runs on Pro and up, with your own API key. We have not run TSK-1, our app-building test, on it yet.
- Status
- Own key
- Plan
- Pro and up, with your own API key (BYOK)
- Where to pick it
- Agent settings
- Your own key
- Google Vertex AI, Google Gemini API
Your own key (BYOK, bring your own key) means you connect your own account with an AI provider, and that provider bills you for the model use.
- Context window
- 1.05MtokensHow much it can read at once
- Max output
- 66KtokensLongest answer it can write
- Released
- Knowledge cutoff
- Newest events it learned about
- Where it runs
- 5servicesOther services that serve it
Published scores
Benchmark scores that Google and test sites published for Gemini 3.1 Pro. Taskade did not re-run them. Lab-reported marks a score from Google itself.
| Benchmark | Score | Source |
|---|---|---|
| SWE-Bench Proresolve rate · publicScale AI | 46.1% | Scale AI |
| Terminal-Benchfirst try · v2.1 · high effort · in Gemini CLIArtificial Analysis | 68.3% | Artificial Analysis |
| OSWorld-Verifiedsuccess rate · Google DeepMindLab-reported | 76.2% | Google DeepMindLab-reported |
| MCP Atlassuccess rate · Google DeepMindLab-reported | 78.2% | Google DeepMindLab-reported |
| GPQA Diamondaccuracy · OpenAI | 94.3% | OpenAI |
| Humanity's Last Examaccuracy · full set, text + MM · Google DeepMindLab-reported | 44.4% | Google DeepMindLab-reported |
| ARC-AGI-2accuracy · Google DeepMindLab-reported | 77.1% | Google DeepMindLab-reported |
| MMMU Proaccuracy · no tools · Google DeepMindLab-reported | 80.5% | Google DeepMindLab-reported |
More published scores (10)
| Benchmark | Score | Source |
|---|---|---|
| SWE-Bench Proresolve rate · Anthropic | 54.2% | Anthropic |
| Terminal-Benchsuccess rate · v2.1 · in Terminus-2 · Anthropic | 70.3% | Anthropic |
| SWE-Atlas Codebase QnAin Mini-SWE-AgentScale AI | 13.5 | Scale AI |
| SWE-Atlas Refactoringin Gemini CLIScale AI | 33.81 | Scale AI |
| SWE-Atlas Test Writingin Mini-SWE-AgentScale AI | 29.84 | Scale AI |
| Artificial Analysis Coding Agent Indexaverage, first try · high effort · in Gemini CLIArtificial Analysis | 43% | Artificial Analysis |
| SWE-Atlas Codebase QnAfirst try · high effort · in Gemini CLIArtificial Analysis | 45.6% | Artificial Analysis |
| SWE-Bench Profirst try · high effort · in Gemini CLI · hard-aaArtificial Analysis | 15.1% | Artificial Analysis |
| CharXiv Reasoningaccuracy · no tools · Google DeepMindLab-reported | 83.3% | Google DeepMindLab-reported |
| GDPval-AAElo rating · Google DeepMindLab-reported | 1,314 | Google DeepMindLab-reported |
LMArena scores
| Benchmark | Score | Source |
|---|---|---|
| Text arena±3 · 132,547 votes · Rank 20 of 419 · LMArena | 1,487 | LMArena |
| WebDev arena±5 · 24,129 votes · Rank 80 of 152 · LMArena | 1,447 | LMArena |
Scores from the LMArena leaderboard dataset (CC BY 4.0). People vote between two answers, and the votes set the score. Taskade picked the Overall category and the default setting of each model.
Scores collected by models.dev (MIT). Each source links the page that published the score.
Versions tested
Not tested yet
Model facts
Published specs from the model maker. A token is about three quarters of a word. Taskade can run the model with lower limits than these.
- Last updated
- Reads
- Text, Images, Video, Audio, PDFWhat you can give it
- Writes
- TextWhat it gives back
- Reasoning
- YesThinking a problem through first
- Tool calling
- YesUsing tools, such as web search
- File attachments
- YesReading files that you attach
- Structured output
- YesFilling a set format, such as a table
- Open weights
- NoFree for anyone to download and run
- Thinking levels
- LowMediumHighHow long it can think. Higher is slower and more careful.
Published specs from the model maker, via models.dev (MIT).
Run Gemini 3.1 Pro with your own key
Pick it with your own key in a custom agent.
| Key provider | Plan |
|---|---|
| Google Vertex AI | Pro and up |
| Google Gemini API | Pro and up |
Where it runs
Services outside Taskade that serve Gemini 3.1 Pro, from models.dev. Each link opens in a new tab. In Taskade, your own key works only through the key providers on this page.
How to choose Gemini 3.1 Pro
- Open the agent settings.
- Under AI Model, choose the model.
- Under Bring Your Own Key, add or select your provider key.
Compare Gemini
Other Google models
FAQ
Which Taskade plans include Gemini 3.1 Pro?
Gemini 3.1 Pro runs on Pro and up, with your own API key. Your provider bills the model use to your own account.
Has Taskade tested Gemini 3.1 Pro?
Not yet. Gemini 3.1 Pro has no TSK-1 app test yet. The Gemini results page shows how other Google models build apps.
How do I choose Gemini 3.1 Pro in Taskade?
Open the agent settings. Under AI Model, choose Gemini 3.1 Pro. Under Bring Your Own Key, select your provider key.
Can I use my own API key for Gemini 3.1 Pro?
Yes. In a custom agent, you can pick Gemini 3.1 Pro with your own key from Google Vertex AI or Google Gemini API (Pro and up). Your provider bills the model use to your own account.
Who makes Gemini 3.1 Pro?
Google makes Gemini 3.1 Pro. Taskade runs it on Pro and up, with your own API key, beside frontier models from other top AI labs.