Gemini 3.1 Pro

Frontier

Google•Released on 2026-02-19

Google's most advanced Pro-tier model with 1M context, dynamic thinking, and the highest ARC-AGI-2 score (77.1%) among all models. Excels at multimodal reasoning across text, images, audio, and video. Best price-to-performance ratio among frontier models.

86
Overall Score

Core Specs

1000K
Context Window
66K
Max Output
✓
Reasoning
✗
Open Source
Multimodal Support
textimageaudiovideo

Scenario Scores

Coding
#10 / 1485
Writing
#6 / 1486
Translation
#6 / 1488
Data Analysis
#6 / 1487
Conversation
#7 / 1485
Image Gen
#3 / 684

User Feedback Highlights

Based on community feedback. Hover to see original reviews.

+ Best for frontend/web design+ Cheapest frontier model ($2/$12)− Weaker at agentic tasks− CLI/agent harness quality inconsistent+ 1M context window− Sometimes skips lines in code edits− Context window issues reported+ Free tier via AI Studio+ Native multimodal (text/image/audio/video)+ Highest ARC-AGI-2 score (77.1%)
Sentiment:👍 52%😐 28%👎 20%

Pros & Cons

Pros

  • +Cheapest frontier model ($2/$12)
  • +Highest ARC-AGI-2 score (77.1%)
  • +Native multimodal (text/image/audio/video)
  • +1M context window
  • +Free tier via AI Studio
  • +Best for frontend/web design

Cons

  • −Weaker at agentic tasks
  • −Context window issues reported
  • −Sometimes skips lines in code edits
  • −CLI/agent harness quality inconsistent

Reliability

SLA99.5%
Incidents (30d)3
Last Incident2026-03-02
Gemini 3 Pro 将于 2026-03-09 下线，建议迁移至 3.1 Pro
View Status Page →

Pricing

Input (per 1M tokens)$2.00
Output (per 1M tokens)$12.00
Subscription$20/month
Free trial available
Updated on 2026-03-06

Tools Supporting This Model

Cursor
Anysphere

Compare with Others

Gemini 3.1 Pro vs GPT-5.4 →Gemini 3.1 Pro vs GPT-5.4 Thinking →Gemini 3.1 Pro vs Claude Opus 4.6 →

Benchmarks

arcAgi277.1%
mmlu91.8%
sweBench76.2%
gpqaDiamond91.9%