Google Gemini
## what it does
Overview
Google’s flagship AI is Gemini 3.8 Flash (announced Sep 2026 on the DeepMind blog) — pitched as “best for tackling complex agentic tasks at scale.” The DeepMind models page (fetched 2026-09-10) also ships Gemini 3.8 Flash Cyber (defenders / Fairwind), Gemini 3.7 Flash, Gemini 3.5 Transcribe, Gemini Robotics 2, Lyria 3.5, Nano Banana 2 Lite (image), and Gemma 4 (open weights). Consumer access is via the Gemini app, Google AI Studio, Google Antigravity (Google’s agentic IDE), and the Gemini Enterprise Agent Platform.
Key Strengths
| Capability | Rating | Notes |
|---|---|---|
| Workspace integration | ⭐⭐⭐⭐⭐ | Native in Gmail, Docs, Sheets, Slides |
| Multimodal | ⭐⭐⭐⭐⭐ | Image/video/audio via Gemini, Nano Banana, Veo, Lyria |
| Search grounding | ⭐⭐⭐⭐⭐ | Real-time Google Search, citations baked in |
| Agentic dev | ⭐⭐⭐⭐⭐ | Antigravity is Google’s agentic IDE (fetched on DeepMind page) |
| Flash cadence | ⭐⭐⭐⭐⭐ | 3.5/3.6/3.7/3.8 in ~12 months; flash line moves fast |
| Context window | ⭐⭐⭐⭐⭐ | 1M+ tokens remains the largest in industry |
Current model line (Sep 2026)
From deepmind.google/models (fetched 2026-09-10) and the 3.8 Flash launch post:
| Model | Role | Notes |
|---|---|---|
| Gemini 3.8 Flash | Current flash flagship | ”Best for tackling complex agentic tasks at scale.” Sep 2026 launch. |
| Gemini 3.8 Flash Cyber | Defenders / Fairwind | Sep 2026 launch |
| Gemini 3.7 Flash | Workhorse for coding/agents | Aug 2026 launch |
| Gemini 3.5 Transcribe | Speech-to-text | Sep 2026 launch |
| Gemini Robotics 2 | Embodied agents | ”Intelligence layer for any kind of robot” |
| Gemini Omni 1.1 Flash | Creative control | Aug 2026 launch |
| Nano Banana 2 Lite | Image gen | ”Fastest, most efficient Gemini Image model” |
| Lyria 3.5 | Music | ”Improved vocals and acoustic details” |
| Gemma 4 | Open weights | ”Most intelligent open models, built to maximize intelligence-per-parameter” |
The consumer access line on Google’s post is Google AI Pro and Ultra.
API pricing (Perplexity-routed rates, fetched 2026-09-10)
Source: docs.perplexity.ai/getting-started/pricing. Per 1M tokens.
| Model | Input | Output | Cache |
|---|---|---|---|
google/gemini-3.7-flash | $0.75 | $3.75 | $0.075 |
google/gemini-3.1-pro-preview | $2 (<200k) / $4 (≥200k) | $12 (<200k) / $18 (≥200k) | input × 0.1 |
google/gemini-3.1-flash-lite | $0.25 | $1.50 | input × 0.1 |
google/gemini-3.6-flash | $1.50 | $7.50 | $0.15 |
google/gemini-3.5-flash | $1.50 | $9.00 | $0.15 |
google/gemini-3.5-flash-lite | $0.30 | $2.50 | $0.03 |
google/gemini-3-flash-preview | $0.50 | $3.00 | input × 0.1 |
best-models-by-lab.md notes 3.8 Flash API intro pricing of $0.75 / $3.75 per 1M tokens until 31 Dec 2026. Use the live Google AI Studio or Vertex AI for Google’s own published price.
Where it ships
| Surface | What you get |
|---|---|
| Gemini app | Consumer chat + image + video + audio |
| Google AI Studio | Browser dev playground |
| Google Antigravity | Agentic development platform (fetched on DeepMind models page) |
| Gemini Enterprise Agent Platform | Build / scale / govern agents |
| Workspace | Gemini in Docs, Gmail, Sheets, Slides |
| Vertex AI | GCP-hosted Gemini + Imagen + Veo |
Gemini for Your Industry
| Industry | Use Case | Link |
|---|---|---|
| Office Work | Email drafting in Gmail, spreadsheet analysis | Gemini for Work |
| Business | Market research with Search grounding, presentations | Gemini for Business |
| Education | Student feedback, curriculum design | Gemini for Education |
Privacy & Data
- Workspace data — Gemini in Workspace doesn’t use your data for training
- API — Data not used for training by default
- Enterprise — Standard Google Cloud compliance (SOC 2, ISO 27001, HIPAA)
Comparisons
Related
- DeepMind models page — Full lineup
- Gemini 3.8 Flash launch — Sep 2026 launch post
## embed this badge
