Google Gemini
Google's assistant and API. Gemini 3.1 Pro carries a 1,048,576 token context on every tier and the lowest frontier input price at $2 per million.
Overview
The case for Gemini is economic. Gemini 3.1 Pro gives you 1,048,576 tokens of context at $2 per million input and $12 output, stepping to $4 and $18 only once a prompt crosses 200K. Batch halves it again. Cached reads land at $0.20 per million. For contract sets, regulatory filings, long transcripts or a full repository read in one pass, nothing else at this quality level is close on cost.
Multimodality is the second reason to look. Audio, video, images and PDFs enter the same request as text, which removes an entire class of preprocessing glue. Deep Think, gated behind the $100 or $200 Google AI Ultra plans, adds longer deliberation for harder reasoning. Distribution is broad: AI Studio for prototyping, Vertex AI for anything needing regional deployment and audit logs, plus Gemini Enterprise and Android Studio.
The problems deserve plain statement. Its successor has slipped three times, which is why the Flash line has run ahead to 3.7 while the flagship sits at 3.1, and that makes roadmap planning awkward. Coding benchmarks put it behind, roughly 80.6% on SWE-bench Verified against GPT-5.5's 88.7%, and multi-step refactoring is where the gap shows. Data policy needs attention too. The paid API and Vertex do not train on customer content, but free AI Studio usage does outside the EEA, Switzerland and the UK, and Google has been moving free and paid users off the open source Gemini CLI onto a closed source replacement with tighter quotas.
Key Features
- ✓ Gemini 3.1 Pro with a 1,048,576 token context window and 65,536 maximum output tokens, released 19 February 2026
- ✓ Tiered pricing that only steps up past 200K input tokens, so most prompts stay at the $2 / $12 rate
- ✓ Native multimodal input covering text, images, video, audio and PDFs in a single request
- ✓ Deep Think mode on the 3.1 line for extended deliberation, gated behind Google AI Ultra
- ✓ Distribution through AI Studio, Vertex AI, Gemini Enterprise, Android Studio and Antigravity
Where it holds
- • Cheapest route to a genuine 1M token window at frontier quality. $2 per million input is under half Sol's rate.
- • Multimodal handling is the least fussy of the big three. Video and PDF go straight in without an extraction stage in front of it.
- • Vertex AI brings regional deployment, IAM and audit logging, which is usually what procurement actually asks about.
Where it breaks
- • The Pro line has stalled. 3.1 Pro shipped in February and 3.5 Pro has missed a late June date, a 17 July date and an early August date, so Google now ships Flash at 3.7 while its flagship sits at 3.1.
- • Instruction following is the recurring developer complaint, and it surfaces in agent loops long before it surfaces in chat. SWE-bench Verified lands near 80.6% against GPT-5.5's roughly 88.7%.
- • Free AI Studio content is used to improve Google's products outside the EEA, Switzerland and the UK. Client data does not belong on that tier.
- • Gemini CLI is being folded into the closed source Antigravity CLI, with free quotas tight enough that developers report hitting weekly caps within a handful of turns.
My Take
Against Sol it loses on agentic reliability and wins on price, which for document-heavy workloads is usually the trade you want. $2 per million input up to 200K on a real 1,048,576 token window, with video and PDFs going in without a conversion step. What holds it back is the Pro line stalling: 3.1 Pro shipped 19 February 2026 and 3.5 Pro has now slipped three announced dates, so the Flash line reads 3.7 while the flagship reads 3.1. Instruction drift is a genuine complaint rather than a benchmark artefact, and it bites hardest inside long agent loops.
Quick Info
- Pricing:
- freemium
- Openness:
- Proprietary
- Starting at:
- Free tier in the Gemini app and AI Studio. Google AI Pro $19.99/mo; Google AI Ultra restructured at I/O 2026 in May into a $100/mo developer tier and a $200/mo tier, down from the previous single $250 price. API per million tokens for Gemini 3.1 Pro: $2 input / $12 output up to 200K, rising to $4 / $18 above 200K. Batch API at 50% off, context caching at $0.20 per million cached read tokens under 200K and $0.40 above.
- Added:
- Aug 2026
- Updated:
- Aug 2026
Use Cases
Judge it on your own work
The notes above say where Google Gemini holds and where it breaks. The fastest check is your own workload.
Visit website ↗Alternatives to Google Gemini
ChatGPT
freemiumOpenAI's assistant and API, now on the GPT-5.6 Sol, Terra and Luna tiers, with a 1M token context window on all three and reasoning effort you set per request.
Claude
freemiumAI assistant focused on safety, accuracy, and nuanced understanding for complex tasks and analysis.
DeepSeek
freemiumChinese AI lab producing open-source LLMs that rival proprietary models at a fraction of the cost, with strong reasoning and coding abilities.
Tongyi Qianwen / 通义千问 (Qwen)
freemiumAlibaba’s open-source large language model family and AI assistant, offering strong multilingual capabilities and deep enterprise integration with Alibaba Cloud.
Mistral AI
open-sourceEuropean lab publishing open weight models you can run on your own hardware. Large 3 is a 675B MoE with 41B active, 256K context and Apache 2.0 weights.