Gemini 3.1 Flash-Lite and Workspace AI: Pricing, Rollout, and What to Do Next (March 2026)
Gemini 3.1 Flash-Lite (March 2026) is Google’s preview Gemini 3–series API model positioned for high-volume, latency- and cost-sensitive workloads; Workspace AI updates in the same window push Gemini deeper into Docs, Sheets, Slides, and Drive for Google AI Ultra and Pro subscribers, with English-first rollout and region limits on some Drive features.
TL;DR: According to Google’s March 2026 announcement, Gemini 3.1 Flash-Lite is the Gemini 3–series option Google positions as fastest and most cost-efficient on the API for developers—listed at $0.25 / 1M input tokens and $1.50 / 1M output tokens, in preview through Google AI Studio and Vertex AI. On a separate track, Gemini in Docs, Sheets, Slides, and Drive is gaining beta capabilities for Google AI Ultra and Pro subscribers: English for Docs, Sheets, and Slides globally; Drive-related features first in the U.S., with more languages and product polish planned.
If you run high-volume LLM backends or spend your day in Workspace, both tracks matter: lower per-token cost for apps and APIs, and deeper Gemini inside the files teams already share. For surrounding news, see Top 10 AI and Tech Stories This Week (March 17–24, 2026).
At a glance (March 2026)
- Gemini 3.1 Flash-Lite (API): Preview SKU; $0.25/1M input, $1.50/1M output per Google’s public pricing (Google Keyword, Mar 3, 2026).
- Where to run it: Google AI Studio / Gemini API and Vertex AI (preview model IDs and quotas apply).
- Workspace Gemini: Beta features for Google AI Ultra and Pro; Docs/Sheets/Slides in English worldwide; Drive rollout begins U.S.-first (Google Keyword, Mar 10, 2026).
- Caveat: Preview and beta labels mean IDs, limits, regions, and pricing can change before general availability—validate in-console before you commit architecture.
What changed (two tracks)
Google announced two related but distinct updates:
- Developer / cloud API — Gemini 3.1 Flash-Lite preview for the Gemini API (AI Studio) and Vertex AI, aimed at latency- and cost-sensitive workloads at scale (Google Keyword, Mar 3, 2026).
- Workspace — Broader Gemini in Docs, Sheets, Slides, and Drive in beta for Google AI Ultra and Pro, including spreadsheet generation/editing, richer Docs drafting, Slides layout assistance, and Drive Q&A framed as “Ask Gemini”–style help (Google Keyword, Mar 10, 2026).
Pricing and how Google positions performance
API pricing (as published by Google): $0.25 per 1M input tokens and $1.50 per 1M output tokens for Gemini 3.1 Flash-Lite (Google Keyword). In its own messaging, Google describes the model as the fastest and most cost-efficient in the Gemini 3 line for high-volume developer workloads.
Latency, throughput, and leaderboard figures (all Google-sourced): Against Gemini 2.5 Flash, Google reports roughly 2.5× faster time to first answer token and about 45% higher output speed, while claiming similar or better quality, citing Artificial Analysis as the measurement context (Google Keyword). For Arena.ai Elo, Google cites 1432 for 3.1 Flash-Lite (Google Keyword). Treat these as vendor-reported orientation, not a substitute for tests on your prompts and data.
Model card (DeepMind): Documented inputs include text, image, audio, and video; up to ~1M-token context and up to 64K-token output (Google DeepMind model card). Thinking levels in AI Studio and Vertex AI adjust how much “reasoning” the model applies per task (Google Keyword).
Workspace (Sheets): Google describes Gemini in Sheets as state-of-the-art on the SpreadsheetBench benchmark (70.48% success) and says it beats named competitors and nears human expert performance on that benchmark (Google Keyword). That is a dataset-specific claim, not a guarantee for every spreadsheet task.
Rollout caveats (read before you rebuild on it)
- Preview, not GA — 3.1 Flash-Lite is a preview SKU in AI Studio and Vertex AI; names, limits, and defaults can change before general availability (Google Keyword).
- Workspace is beta and tier-gated — New Docs/Sheets/Slides/Drive experiences are beta and prioritized for Google AI Ultra and Pro; Google says it will refine the product and expand languages over time (Google Keyword).
- Regional coverage — Docs, Sheets, Slides: English, available globally per Google’s March 2026 post. Drive: features described there (including Ask Gemini / AI Overviews-style experiences) start in the United States (Google Keyword).
- Sheets “Fill with Gemini” — Google cites a 95-participant study comparing manual entry with Fill with Gemini on a 100-cell task (Google Keyword).
- Slides — Full deck generation from one prompt is explicitly coming soon (Google Keyword).
- Benchmarks ≠ your stack — Leaderboard and benchmark numbers help orient; your prompts, tools, retrieval, and evals still decide production quality.
Practical implications for builders
- Cost-sensitive pipelines — Classification, routing, moderation, translation, and high-QPS chat backends are natural fits when $/token and latency dominate. Pair structured web data from Apify with Flash-Lite for enrichment or summarization at scale. Patterns: web scraping with AI and LLMs in 2026.
- Multimodal inputs — If you process images, audio, or video with text, Flash-Lite’s documented multimodal inputs support single-model pipelines (Google DeepMind model card).
- Workspace vs API — Workspace Gemini is built around files, email, and collaboration inside Google’s apps. Gemini API / Vertex is for your backends and products. Same brand family; not interchangeable deployment paths.
- Orchestration — For multi-step automation (CRM updates, scrape → transform → notify), Make.com is a common glue layer; see Make.com and Apify for web scraping and MCP servers for web scraping for tool-native agent setups.
Vendor positioning snapshot (API tier, Google-reported figures)
The table below copies selected rows from Google DeepMind’s Gemini 3.1 Flash-Lite model card (March 2026): listed API prices (no caching) and output throughput for vendor-labeled models in that document. It reflects Google’s table, not independent price verification.
| Model (label in Google DeepMind table) | Input $/1M tokens | Output $/1M tokens | Output tokens/s |
|---|---|---|---|
| Gemini 3.1 Flash-Lite (High) | $0.25 | $1.50 | 363 |
| Gemini 2.5 Flash (Dynamic) | $0.30 | $2.50 | 249 |
| Gemini 2.5 Flash-Lite (Dynamic) | $0.10 | $0.40 | 366 |
| GPT-5 mini (High) | $0.25 | $2.00 | 71 |
| Claude Haiku 4.5 (Extended Thinking) | $1.00 | $5.00 | 108 |
| Grok 4.1 Fast (Reasoning) | $0.20 | $0.50 | 145 |
Source: Google DeepMind — Gemini 3.1 Flash-Lite model card (March 2026). Confirm current rates and model IDs on each provider’s pricing page before budgeting.
Action checklist
- Confirm SKUs — In AI Studio / Vertex, record the exact preview model ID and any quota or region limits before load testing.
- Re-run your evals — Compare Flash-Lite with your current default model on your tasks (extraction, support replies, routing)—not only on public benchmarks.
- Separate cost models — Model $/1M input vs output against your token mix; cheap input pricing still hurts when output volume is high.
- Workspace pilot — If you have Google AI Ultra/Pro, pilot Docs/Sheets on non-sensitive work first; Drive capabilities may be unavailable outside the initial geography.
- Data boundaries — For regulated data, align with admin policies, DLP, and retention before turning on file- and email-grounded Gemini features.
- Automation path — When APIs beat in-app assistants for your use case, keep extraction in tools like Apify and LLM steps in Vertex or your app; for doc-to-LLM ingestion, Firecrawl can supply clean markdown.
Weekly context: Top 10 AI and Tech Stories This Week (March 17–24, 2026). Pipelines: LangChain Apify content pipeline. Agents: AI agent frameworks in 2026. Data plane: Apify Store · Free Apify signup.
Related Deep Dives This Week
Google lists $0.25 per 1M input tokens and $1.50 per 1M output tokens for Gemini 3.1 Flash-Lite (Google Keyword, March 2026). Confirm current rates in Google AI Studio, the Gemini API documentation, or Vertex AI billing before you budget.
Google announced preview access through the Gemini API in Google AI Studio and for enterprises through Vertex AI (Google Keyword, March 2026). Use the preview model ID shown in those consoles.
Google describes the updates as beta for Google AI Ultra and Pro subscribers: English globally for Docs, Sheets, and Slides, with Drive-related features starting in the United States (Google Keyword, March 2026).
They share the Gemini brand and product direction, but Workspace features ship inside Google Docs, Sheets, Slides, and Drive with subscription, beta, and region constraints. The Gemini API is for applications and backends you control.
Preview SKUs can change; vendor benchmarks do not guarantee results on your data; Workspace betas may shift by language, region, and plan. Run private evaluations and review policy, safety, and compliance requirements.