Gemini 3 Models by Google DeepMind
Google DeepMindProprietary
6 models2025–2026Up to 1M ctxFrom $0.5/1M input
About
Google DeepMind's Gemini 3 model family. Features frontier-class intelligence across Pro, Flash, and Lite variants with up to 1M token context windows.
Current Variants
Use-when guidance is derived from seed capabilities, context, release, and replacement fields.
5 in view1 retired
Use when the workload needs robotics, 128K context, and reasoning.
2026-04robotics128K contextreasoning
Gemini 3 Flash PreviewCurrent
Use when the workload needs 1M context, tool use, and function calling.
2025-121M contexttool usefunction calling
Gemini 3 FlashCurrent
Use when the workload needs 1M context, tool use, and function calling.
2025-121M contexttool usefunction calling
Gemini 3 ProCurrent
Use when the workload needs 1M context, tool use, and function calling.
2025-121M contexttool usefunction calling
Use when the workload needs 66K context.
2025-0966K context
| Model | Use when | Released | Signals | Status |
|---|---|---|---|---|
| Gemini Robotics-ER 1.6 Preview | Use when the workload needs robotics, 128K context, and reasoning. | 2026-04 | robotics128K contextreasoning | Current |
| Gemini 3 Flash Preview | Use when the workload needs 1M context, tool use, and function calling. | 2025-12 | 1M contexttool usefunction calling | Current |
| Gemini 3 Flash | Use when the workload needs 1M context, tool use, and function calling. | 2025-12 | 1M contexttool usefunction calling | Current |
| Gemini 3 Pro | Use when the workload needs 1M context, tool use, and function calling. | 2025-12 | 1M contexttool usefunction calling | Current |
| Nano Banana Pro (Gemini 3 Pro Image Preview) | Use when the workload needs 66K context. | 2025-09 | 66K context | Current |
Release Timeline
3 release groups2026-04
1 current
Gemini Robotics-ER 1.6 Preview
Currentrobotics128K contextreasoning
2025-12
3 current
Gemini 3 Flash
Current1M contexttool usefunction calling
Gemini 3 Flash Preview
Current1M contexttool usefunction calling
Gemini 3 Pro
Current1M contexttool usefunction calling
2025-09
1 current · 1 retired
Gemini 3 Pro Preview
Archived1M contexttool usefunction calling
Nano Banana Pro (Gemini 3 Pro Image Preview)
Current66K context
Specifications(6 models)
| Model | Released | Context | Vision | Multimodal | Reasoning | Fn Calling | Tool Use | Structured Outputs | Code Exec |
|---|---|---|---|---|---|---|---|---|---|
| Gemini Robotics-ER 1.6 Preview | 2026-04 | 128K | Yes | Yes | Yes | Yes | Yes | Yes | No |
| Gemini 3 Flash Preview | 2025-12 | 1M | Yes | Yes | No | Yes | Yes | Yes | Yes |
| Gemini 3 Flash | 2025-12 | 1M | Yes | Yes | No | Yes | Yes | No | Yes |
| Gemini 3 Pro | 2025-12 | 1M | Yes | Yes | No | Yes | Yes | No | Yes |
| Nano Banana Pro (Gemini 3 Pro Image Preview) | 2025-09 | 66K | No | No | No | No | No | No | No |
Available From(4 providers)
Pricing
| Model | Provider | Input / 1M | Output / 1M | Type |
|---|---|---|---|---|
| Gemini 3 Flash Preview | GCP Vertex AI | $0.5 | $3 | Serverless |
| Gemini 3 Flash Preview | OpenRouter | $0.5 | $3 | Serverless |
| Gemini 3 Flash Preview | Google AI Studio | $0.5 | $3 | Serverless |
| Gemini 3 Flash | Replicate API | $0.5 | $3 | Serverless |
| Gemini 3 Flash | GCP Vertex AI | $0.5 | $3 | Serverless |
| Gemini 3 Flash | Google AI Studio | $0.5 | $3 | Serverless |
| Gemini 3 Pro | GCP Vertex AI | $1.25 | $5 | Serverless |
| Nano Banana Pro (Gemini 3 Pro Image Preview) | Google AI Studio | $2 | $120 | Serverless |
| Gemini 3 Pro | Replicate API | $2 | $12 | Serverless |
| Nano Banana Pro (Gemini 3 Pro Image Preview) | OpenRouter | $2 | $12 | Serverless |
| Nano Banana Pro (Gemini 3 Pro Image Preview) | GCP Vertex AI | $3 | $15 | Serverless |
Frequently Asked Questions
- What is Gemini 3 used for?
- Gemini 3 is used for robotics, vision and multimodal work, and reasoning. The family description and listed model capabilities point to those workloads as the best fit.
- How does Gemini 3 compare to Gemma 4?
- Gemini 3 by Google DeepMind is strongest where you need robotics, while Gemma 4 by Google DeepMind is the closest related family to check for vision and multimodal work. Gemini 3 has 6 listed variants and reaches up to 1M context, while Gemma 4 reaches up to 256K context, so compare the specs and pricing tables before choosing a production model.
- Which Gemini 3 model should I use?
- For the lowest listed input price, start with Gemini 3 Flash Preview through Google AI Studio at $0.5/1M input tokens. For the most capable/latest local choice, evaluate Gemini 3 Flash Preview with 1M context and tool use, function calling, structured outputs, and multimodal inputs.
