Gemini app (Gemini 3.1 Pro and 3.5 Flash)
8/10The consumer-facing Gemini chat app on web, iOS and Android, now running Gemini 3.1 Pro for reasoning and Gemini 3.5 Flash for fast everyday tasks. Following I/O 2026 it has a new Neural Expressive interface, a unified Tools menu and a Daily Brief agent that prepares a prioritised digest of your day. Sits behind Google AI Plus, Pro and Ultra subscriptions, with a capped free tier.
Key features
Gemini 3.1 Pro for hard reasoning (91.9% GPQA Diamond, 77.1% ARC-AGI-2, 1M token context)Deep Research with collaborative planning, MCP server integration and File SearchDaily Brief agent that prioritises and suggests next steps each morning (Plus, Pro, Ultra)Native multimodal input: image, audio, video, YouTube links, screen shareGemini Spark personal agent on Ultra, running 24/7 on Google Cloud VMs (US only at launch)
Best use cases
- Summarise a 90-minute YouTube video into action points before a supplier meeting
- Run a Deep Research brief on a target acquisition with traceable citations
- Convert a PDF contractor quote and a site photo into a structured comparison table
- Use Daily Brief to prep morning standups from calendar plus inbox
- Translate and analyse a recorded customer call in any major European language
Weaknesses
Still produces more generic, lightly hedged prose than Claude. Spark agent is US-only for now. Long-context performance is strong on retrieval but weaker on multi-step writing instructions than GPT-5.1.
Pricing
Free tier (capped). Google AI Plus £6.99/month. Google AI Pro £18.99/month. Google AI Ultra £79.99/month (cut from £189.99 in May 2026 for the standard tier; higher allotment tier still available).
Reach for this when the task involves video, YouTube, deep research with citations, or anything sitting inside your Google account.
Gemini in Google Workspace
9/10Gemini built directly into Gmail, Docs, Sheets, Slides, Meet and Drive. The old add-on SKU is gone: as of 2026 every paid Workspace plan includes Gemini, with the deeper Docs and Sheets features starting at Business Standard. This is the version most UK SMEs will actually use day to day.
Key features
Help me write in Gmail and Docs, with the company's own Drive content as groundingSheets formula generation, data cleaning and chart suggestions from natural languageMeet note-taking, transcription and action item extraction in real timeSide panel chat in every app that can reference the document, thread or sheet open in front of youEnterprise data residency and admin controls; content not used to train models
Best use cases
- Draft a reply to a long supplier email chain with the right tone and history
- Generate first-draft minutes plus owners and dates from a Meet recording
- Build a Sheets margin model from a verbal brief instead of fiddling with INDEX MATCH
- Turn a Drive folder of policy PDFs into a Q&A side panel for the operations team
- Produce a Slides client update deck from a Doc in one prompt
Weaknesses
Quality is uneven across surfaces; Meet notes and Gmail are strong, Slides generation is still rough. Business Starter only gets Gmail and the Gemini app, not the full suite. If your team is half on Microsoft 365, you are paying for both Copilot and Gemini.
Pricing
Included in Workspace plans: Business Starter from £6.65/user/month (limited Gemini); Business Standard £13.20/user/month (full Gemini in Docs, Sheets, Meet); Business Plus £21.10/user/month. Pricing rises modestly from 14 September 2026.
Reach for this if your business runs on Workspace; it is the highest ROI AI per pound for a small ops team.
Google's coding assistant for VS Code, JetBrains, Android Studio and Cloud Workstations. Runs on Gemini 2.5 by default with Gemini 3 Pro available in Enterprise preview. Free for individual developers since March 2026, with paid tiers for teams and tight Google Cloud and BigQuery integration.
Key features
1 million token context window for whole-repo reasoningAgent mode with MCP support and custom commandsFinish Changes feature that completes a multi-file edit task end to endDeep BigQuery, Apigee and Application Integration toolingFree tier: 180,000 completions and 240 chats per day per developer
Best use cases
- Generate a BigQuery SQL pipeline from a plain English brief
- Refactor a multi-file Python service while preserving tests
- Stand up an internal Android app inside Android Studio with agent mode
- Write Terraform for a Google Cloud deployment without leaving the IDE
- Ask whole-codebase questions before onboarding a new contractor
Weaknesses
Tighter GitHub integration sits with Copilot; if your team lives on GitHub you will miss PR summaries and review hooks. Agent mode is improving but still trails Claude Code and Cursor on long-running autonomous tasks. Real ceiling on free tier completions if you code daily.
Pricing
Free for individuals (180k completions/month, 240 chats/day). Standard $19/user/month. Enterprise $45/user/month (Gemini 3 Pro preview, audit logs, private code customisation).
Reach for this if you build on Google Cloud or BigQuery; otherwise Claude Code or Copilot will serve you better.
Imagen 4 (image generation)
7/10Google's flagship text-to-image model, accessible inside the Gemini app, Google AI Studio, Vertex AI and Flow. Strong on photorealistic compositions, accurate text rendering inside images and brand-safe content controls. Sits behind a SynthID watermark by default.
Key features
Accurate in-image text (logos, signage, product labels)Multiple aspect ratios and resolutions up to 2KStyle consistency across a batch via reference imagesSynthID watermarking on every outputAvailable in the Gemini app, AI Studio and Vertex AI for batch and API use
Best use cases
- Mock up a product packaging concept with legible copy on the label
- Generate consistent thumbnails across a 12-video YouTube series
- Produce hero images for a landing page using a brand colour palette
- Visualise a refurb concept from a site photo plus a brief
- Build a library of stock-style team and office images that do not exist on Google Images
Weaknesses
Still trails Midjourney v7 and Higgsfield on cinematic flair and texture. Less expressive on human anatomy and hands than Flux 1.1. Heavier safety filter than other vendors, which trips frequently on legitimate retail and property briefs.
Pricing
Included in Google AI Plus (£6.99/mo), Pro (£18.99/mo) and Ultra (£79.99/mo) with rising quotas. API on Vertex AI: roughly $0.04 per 1024x1024 image.
Reach for this when you need legible text in the image, brand-safe outputs, or batch generation tied to a Google Cloud workflow.
Veo 3.1 and Gemini Omni (video)
9/10Google's flagship text-to-video and image-to-video model, plus the new Gemini Omni series that takes image, audio, video and text in and produces edited video out conversationally. Veo 3.1 generates 8-second clips at up to 4K with synchronised native audio (dialogue, sound effects, ambient). Accessible through the Gemini app, Flow, AI Studio and Vertex AI.
Key features
Native synchronised audio: dialogue, sound effects, ambient, all in one passUp to 4K resolution; upscale to 1080p and 4K available in Flow and APIIngredients to Video: combine reference images of a character, prop and settingVeo 3.1 Lite for cheaper batch generationFlow editor with shot extension, scene chaining and camera direction
Best use cases
- Produce a 30-second product reveal video with synchronised voiceover for social
- Generate consistent character footage across a multi-shot training video
- Mock up a TV-style customer testimonial concept before commissioning a real shoot
- Create animated explainers from a written script with native sound
- B-roll generation for an internal town hall video
Weaknesses
Eight-second hard cap per clip means longer content needs stitching. Still produces uncanny faces in close-ups; not yet usable for any output that needs the same human across many shots without obvious drift. Cost adds up fast on 4K. Heavily safety-filtered.
Pricing
Included in Google AI Pro (£18.99/mo) with a monthly clip allowance and in Ultra (£79.99/mo) with significantly higher limits. API on Vertex AI: Veo 3.1 around $0.40 per second of audio-on output; Veo 3.1 Lite materially cheaper.
Reach for this when you need short-form video with sound built in; the closest thing to a usable AI video model for UK SME marketing in 2026.