Gemini API0 Mixed

GILD Review

GILD's call and the buying facts behind it.

From GILD research

Should we adopt Gemini API?

What will the Gemini API cost your product after Flash's introductory price ends on December 31, 2026?

30-Second TakePick the Gemini API for high-volume or multimodal features, where Flash pricing draws the most praise.
Best fitTeams building multimodal or high-volume features who run their own evaluations and can absorb model upgrades.GILD Review
Consider another workflowIf the goal is coding help for your own engineers, adopt a finished coding agent and skip building one on the API.GILD Review

Drawn from 220 sources and 11 operator insights

GILD analysis

Gemini API Review

Build on the Gemini API for high-volume and multimodal work, where Flash pricing and native document, image, audio, and video input draw the most praise in public accounts, and where DoorDash and Wealthsimple run it beside other providers. Surprise bills lead the concerns, many from old API keys that gained Gemini access, then model retirements and task speed: PartForge moved its default off Gemini 3.8 Flash on October 20, 2026. Flash-Lite prices have risen each generation, pre-release models can shut down two weeks after notice, and spend caps require setup.

AI Model APIs
AI Costs

The decision

Best fit: Teams building multimodal or high-volume features who run their own evaluations and can absorb model upgrades.
Consider another workflow: If the goal is coding help for your own engineers, adopt a finished coding agent and skip building one on the API.
Adoption condition: Set a project spend cap below your tier limit. Use paid-tier billing for customer data. Name an owner for model retirement dates.

Basis: Google documentation, pricing, and terms, the independent Artificial Analysis Intelligence Index, and 194 public sources from engineers and teams, including a DoorDash conference talk, a Pydantic podcast episode, 14 Reddit posts and comments, and 9 GitHub issues and release notes. Three are independent long-context benchmarks, and 12 praise Flash speed on their own workloads.

Buying & deployment factsWhat a team building on it should verify before production traffic grows.

How you buy it

US$0.75 in, US$3.75 out

Per million tokens for Gemini 3.8 Flash through 2026, then US$1.50 and US$7.50.

Source 1, Google, Pricing: Pricing Source 2, Google, Billing: Billing

Details

Usage based, in US dollars per million input and output tokens. Gemini 4 Argon costs US$2 and US$10 through November 30, 2026, then US$4 and US$20. Its cached input costs US$0.10, then US$0.20, with implicit caching only and no free tier. Gemini 3.8 Flash costs US$0.75 and US$3.75 through December 31, 2026, then US$1.50 and US$7.50. Gemini 3.5 Flash-Lite costs US$0.30 and US$2.50, and the pre-release Gemini 3.1 Pro US$2 and US$12 for prompts up to 200,000 tokens. New AI Studio accounts prepay credits from US$5.

Cost drivers

Half price

With Batch or Flex. On 3.8 Flash, cache hits cost a tenth of the input price.

Source 1, Google, Pricing: Pricing

Details

Model choice and output tokens, including thinking, drive cost. Batch and Flex halve the standard rate. On 3.8 Flash, cache hits cost a tenth of the input price. Google Search grounding costs US$14 per 1,000 requests after 5,000 free each month.

Where it runs

Two routes

The Developer API with an AI Studio key, or Google Cloud's Agent Platform, formerly Vertex AI.

Source 2, Google, Billing: Billing Source 5, Google, Gemini Developer API vs. Gemini platform: Gemini Developer API vs. Gemini platform Source 7, cloud.google.com, Introducing Gemini Enterprise Agent Platform (Apr 22, 2026): Introducing Gemini Enterprise Agent Platform (Apr 22, 2026)Source 6, Google, Agent Platform pricing: Agent Platform pricing

Details

Teams call the Gemini Developer API with an AI Studio key or use Google Cloud's Gemini Enterprise Agent Platform, formerly Vertex AI. Global-endpoint prices match. The Agent Platform adds service accounts and Provisioned Throughput, with non-global endpoints priced 10% higher.

What to verify

US$250 a month

The Tier 1 cap, US$2,000 on Tier 2. Free-tier content improves Google's products.

Source 2, Google, Billing: Billing Source 1, Google, Pricing: Pricing Source 3, Google, Additional Terms of Service: Additional Terms of Service Source 4, docs.cloud.google.com, Agent Platform model versions: Agent Platform model versions Source 8, ai.google.dev, Long decode guide: Long decode guide

Details

Monthly caps are US$250 on Tier 1 and US$2,000 on Tier 2. Google uses free-tier content to improve its products. Model retirement dates vary by route. Gemini 2.5 Pro, Flash, and Flash-Lite retired on the Agent Platform on October 20, 2026. Google's Gemini 4 Argon guide gives it one 1,048,576-token window across input, thinking, and output. One hop returns up to 65,536 output tokens, or 32,768 on a non-streaming REST request, and current Google GenAI SDKs continue past the cap automatically.