Best for Coding
Google’s stable Gemini 2.5 Pro supports complex reasoning and coding with a 1M-token input limit. Direct API access is restricted to prior users.
At a glance
API model ID: gemini-2.5-pro
Capabilities & boundaries
Tool support requires the appropriate API integration; a supported tool is not automatically active in every chat.
Thinking is dynamic by default and cannot be disabled. The explicit thinking budget range is 128–32,768 tokens; -1 selects dynamic thinking. These budgets are not the same as Gemini 3 low/medium/high levels.
Put it to work
Original examples from EZ Ai Assist. Adapt these to your task and the features available in your workspace.
Workflow 01
Provide the existing system and constraints before asking for a plan.
Review this database migration proposal against the current schema, traffic constraints, and rollback requirements I provide. Identify data-loss risks and incompatible assumptions. Produce a staged rollout with validation queries and explicit stop conditions. Separate verified constraints from questions for the engineering team.
Workflow 02
Require evidence rather than an invented reconciliation.
Compare these two versions of our technical specification. List changed requirements, contradictions, and downstream test impacts, citing section headings in both documents. Do not silently choose a winner when the documents disagree. Finish with the decisions an owner must make before implementation can begin.
Workflow 03
Ask for falsifiable reasoning and focused tests.
Analyze the algorithm and failing examples below. State its intended invariant, trace where the supplied example violates it, and propose the smallest correction. Include complexity implications and three tests that would distinguish the fix from the old behavior. Do not rewrite unrelated parts of the program.
Developer reference
These are Google API reference prices, not EZ Ai Assist subscription prices.
View EZ Ai Assist plans| Token type | ≤ 200,000input tokens | > 200,000input tokens |
|---|---|---|
| Input | $1.25 | $2.50 |
| Cached input | $0.125 | $0.25 |
| Output (including thinking) | $10.00 | $15.00 |
| Usage | Rate and unit |
|---|---|
| Cache storage | $4.50per 1M tokens per hour |
| Google Search grounding (after allowance) | $35.00per 1,000 grounded prompts |
| Google Maps grounding (after allowance) | $25.00per 1,000 grounded prompts |
Common questions
Consider it for complex code analysis and long-document synthesis. Supply clear constraints and assess correctness, latency, and total usage on your own examples before making it a default.
Thinking is dynamic by default and cannot be disabled. The explicit thinking budget range is 128–32,768 tokens; -1 selects dynamic thinking. These budgets are not the same as Gemini 3 low/medium/high levels.
The Google model card supports these inputs and text output. Upload and tool availability depend on your integration. This does not make the model an image generator, voice generator, or Live API model.
No. Google says the stable 2.5 models remain served, with access limited to users who actively used them before. Retired preview IDs are different. Check current provider and EZ Ai Assist availability independently.
Requests above 200,000 input tokens use higher input, cached-input, and output rates. Cache storage and grounding can add charges. Measure full request usage rather than estimating from your visible prompt alone.
No. Keep evidence relevant, specify the required output, request source references, and check the result. The 1,048,576 input-token limit and 65,536 output-token limit are different constraints.
No. This guide separates Google’s direct API reference from EZ Ai Assist subscriptions. Use the pricing page for plans and the app for current model access. The example prompts are starting points, not guaranteed results.
Check the source
Specifications and API prices checked on . Example prompts and workflow advice are editorial guidance from EZ Ai Assist.
Same provider
Google’s Gemini 3.8 Flash supports multimodal understanding, three thinking levels, and tool-assisted workflows with promotional API pricing.
Google’s Gemini 3.7 Flash combines multimodal inputs, configurable thinking, and structured tool workflows for repeatable everyday tasks.
Google’s Gemini 3.1 Pro Preview supports complex synthesis, coding, and multimodal analysis with a 1M-token input limit.
Historical guide to Google’s retired Gemini 3 Pro Preview API, with specifications, migration checks, and the Gemini 3.1 Pro replacement.