# Gemini Robotics ER 2 — AI Compare Ninja

> Embodied reasoning model for robots via the Gemini API (preview): finds objects, returns points and bounding boxes, plans multi-step tasks and calls your own robot functions.

## Overview
- **Category:** Development & Code (Embodied AI & robotics)
- **Status:** Active · checked
- **Paid from:** $1 / $5 / pay as you go (Paid Tier)
- **Free access:** Free plan
- **Website:** https://ai.google.dev/gemini-api/docs/robotics-overview
- **Canonical URL:** https://www.aicompare.ninja/en/tool/gemini-robotics-er

## What it does
- Embodied reasoning model for robots via the Gemini API (preview): finds objects, returns points and bounding boxes, plans multi-step tasks and calls your own robot functions.
- Free tier at no charge; paid tier $1 / $5 per 1M tokens, time-limited through 2026-12-31, $2 / $10 from 2027-01-01.

## In brief
Embodied reasoning model for robots via the Gemini API (preview): finds objects, returns points and bounding boxes, plans multi-step tasks and calls your own robot functions.

Gemini Robotics ER 2 (Google) is, per the vendor, a vision-language model for robotics built on Gemini 3.5 Flash and is in preview. Endpoints: gemini-robotics-er-2-preview (text, image, video and audio input; 131,072 input tokens, 65,536 output tokens; Batch API) and gemini-robotics-er-2-streaming-preview for real time via the Live API. Documented capabilities: spatial reasoning (pointing, object tracking, bounding boxes, trajectories), task orchestration through your own robot APIs, video progress and multi-robot coordination. Free tier at no charge with rate limits; Google states content there is used to improve its products. Paid tier per 1M tokens: input $1, output $5, time-limited through 2026-12-31; from 2027-01-01 input $2, output $10. Batch $0.50 / $2.50 through 2026-12-31, then $1 / $5. Google notes that models can make mistakes and that a safe environment around the robot is the user's responsibility. Prices in US dollars as listed by the vendor, without currency conversion.


## Limits
Gemini Robotics ER 2 (Google) is, per the vendor, a vision-language model for robotics built on Gemini 3.5 Flash and is in preview. Endpoints: gemini-robotics-er-2-preview (text, image, video and audio input; 131,072 input tokens, 65,536 output tokens; Batch API) and gemini-robotics-er-2-streaming-preview for real time via the Live API. Documented capabilities: spatial reasoning (pointing, object tracking, bounding boxes, trajectories), task orchestration through your own robot APIs, video progress and multi-robot coordination. Free tier at no charge with rate limits; Google states content there is used to improve its products. Paid tier per 1M tokens: input $1, output $5, time-limited through 2026-12-31; from 2027-01-01 input $2, output $10. Batch $0.50 / $2.50 through 2026-12-31, then $1 / $5. Google notes that models can make mistakes and that a safe environment around the robot is the user's responsibility. Prices in US dollars as listed by the vendor, without currency conversion.

## Pricing tiers
| Tier | Price | Period | What you get |
| --- | --- | --- | --- |
| Free Tier | free | free | Input and output free of charge with rate limits; per Google, content is used to improve its products; no context caching. |
| Paid Tier | $1 / $5 | pay as you go | Per 1M tokens, time-limited through 2026-12-31: input $1 (text, image, video, audio), output incl. thinking $5, context caching $0.10. From 2027-01-01: $2 / $10 / $0.20. Batch $0.50 / $2.50 through 2026-12-31, then $1 / $5. Per Google, content is not used to improve its products. |

## Sources
All details regarding plans, limits, and features are based on official vendor documentation and the primary sources verified below.

- https://deepmind.google/models/gemini-robotics/
- https://ai.google.dev/gemini-api/docs/robotics-overview
- https://ai.google.dev/gemini-api/docs/pricing
- https://ai.google.dev/gemini-api/terms

---
*Source: AI Compare Ninja (2026) - https://www.aicompare.ninja/en/tool/gemini-robotics-er*