gemini-media-mcp

Overview Schema Source Related Servers Discussions Admin

Generate image

generate_image

Create images from a text prompt, optionally guided by up to 14 reference images for subjects, products, or style. Save results for editing or video generation.

Instructions

Create images from a text prompt, optionally guided by up to 14 reference images (subjects, products, style). Default model nb2 (Nano Banana 2.1, ~~$0.04 per 1K image); pro for the densest scenes and final renders (~~$0.15); nb2-lite for bulk drafts. Returns saved files with URIs you can pass to edit_image, generate_video (image) or other tools.

Input Schema

TableJSON Schema

Name	Required	Description
`count`	No	Number of variations to generate in parallel (1-4, default 1). Each is billed.
`model`	No	Model ID or alias. nb2 (default: Nano Banana 2.1, fast and cheap), pro (Nano Banana Pro, highest fidelity for dense scenes), nb2-lite (cheapest inputs). See list_models.
`prompt`	Yes	What to create. Describe subject, setting, composition, lighting and style in full sentences. Put any exact text to render in quotes.
`imageSize`	No	Output size: 1K (default), 2K or 4K. Larger costs more. 512 only on the deprecated gemini-3.1-flash-image.
`outputName`	No	Optional base file name (without extension) for the saved file(s).
`aspectRatio`	No	e.g. 1:1, 3:2, 2:3, 4:3, 3:4, 4:5, 16:9, 9:16, 21:9 (nb2 also 1:4, 4:1, 1:8, 8:1). Defaults to 1:1 when there are no reference images; with them the model usually follows the first image's ratio.
`googleSearch`	No	Ground the image in Google Search results (current events, real places, data for infographics).
`approvedCostUsd`	No	Only needed when a call exceeds the configured confirmation threshold: the cost in USD the user approved.
`referenceImages`	No	Optional images to guide the result (subjects, products, characters, style): file paths, gemini-media:// URIs from earlier results, or data: URIs. Up to 14 for nb2/pro.

Output Schema

TableJSON Schema

Name	Required	Description
`cost`	Yes
`text`	No	Commentary the model returned alongside the image(s)
`files`	Yes
`model`	Yes
`warnings`	No
`grounding`	No	Google Search queries the model ran and the pages it used (googleSearch only)

Schema Changelog

Changes observed during successful MCP inspections.

Changed18 schema fields changedv1.1.0Oct 14, 2026

addedInput schema / properties / approvedCostUsd

Added value: +{
+  "description": "Only needed when a call exceeds the configured confirmation threshold: the cost in USD the user approved.",
+  "type": "number"
+}

changedInput schema / properties / aspectRatio / description

Previous value: -"Aspect ratio (1:1, 2:3, 3:2, 3:4, 4:3, 4:5, 5:4, 9:16, 16:9, 21:9)"New value: +"e.g. 1:1, 3:2, 2:3, 4:3, 3:4, 4:5, 16:9, 9:16, 21:9 (nb2 also 1:4, 4:1, 1:8, 8:1). Defaults to 1:1 when there are no reference images; with them the model usually follows the first image's ratio."

addedInput schema / properties / count

Added value: +{
+  "description": "Number of variations to generate in parallel (1-4, default 1). Each is billed.",
+  "type": "integer"
+}

addedInput schema / properties / googleSearch

Added value: +{
+  "description": "Ground the image in Google Search results (current events, real places, data for infographics).",
+  "type": "boolean"
+}

addedInput schema / properties / imageSize

Added value: +{
+  "description": "Output size: 1K (default), 2K or 4K. Larger costs more. 512 only on the deprecated gemini-3.1-flash-image.",
+  "type": "string"
+}

changedInput schema / properties / model / description

Previous value: -"Model tier: nb2 (default) or pro. Raw model IDs also accepted"New value: +"Model ID or alias. nb2 (default: Nano Banana 2.1, fast and cheap), pro (Nano Banana Pro, highest fidelity for dense scenes), nb2-lite (cheapest inputs). See list_models."

addedInput schema / properties / outputName

Added value: +{
+  "description": "Optional base file name (without extension) for the saved file(s).",
+  "type": "string"
+}

changedInput schema / properties / prompt / description

Previous value: -"Text description of the image to generate"New value: +"What to create. Describe subject, setting, composition, lighting and style in full sentences. Put any exact text to render in quotes."

addedInput schema / properties / referenceImages

Added value: +{
+  "description": "Optional images to guide the result (subjects, products, characters, style): file paths, gemini-media:// URIs from earlier results, or data: URIs. Up to 14 for nb2/pro.",
+  "items": {
+    "type": "string"
+  },
+  "type": [
+    "null",
+    "array"
+  ]
+}

removedInput schema / properties / resolution

Removed value: -{
-  "description": "Output resolution (1K, 2K, 4K)",
-  "type": "string"
-}

addedOutput schema / properties / cost

Added value: +{
+  "additionalProperties": false,
+  "properties": {
+    "basis": {
+      "description": "How usd was derived: usage_metadata, unit_params, token_estimate or unpriced",
+      "type": "string"
+    },
+    "breakdown": {
+      "type": "string"
+    },
+    "estimatedUsd": {
+      "description": "Estimate computed before the call from the price table",
+      "type": "number"
+    },
+    "pending": {
+      "description": "True while a video job is still running (cost is reserved, not final)",
+      "type": "boolean"
+    },
+    "priceAsOf": {
+      "type": "string"
+    },
+    "usd": {
+      "description": "Best-known cost in USD; reconciled from token usage when the API reports it",
+      "type": "number"
+    }
+  },
+  "required": [
+    "estimatedUsd",
+    "usd",
+    "basis"
+  ],
+  "type": "object"
+}

removedOutput schema / properties / filePath
```
Removed value: -{
-  "type": "string"
-}
```

addedOutput schema / properties / files

Added value: +{
+  "items": {
+    "additionalProperties": false,
+    "properties": {
+      "bytes": {
+        "type": "integer"
+      },
+      "durationSeconds": {
+        "type": "number"
+      },
+      "height": {
+        "type": "integer"
+      },
+      "mimeType": {
+        "type": "string"
+      },
+      "name": {
+        "type": "string"
+      },
+      "path": {
+        "description": "Absolute path of the saved file on the server's filesystem",
+        "type": "string"
+      },
+      "uri": {
+        "description": "MCP resource URI; pass it back as an input to other tools or read it with resources/read",
+        "type": "string"
+      },
+      "width": {
+        "type": "integer"
+      }
+    },
+    "required": [
+      "path",
+      "uri",
+      "name",
+      "mimeType",
+      "bytes"
+    ],
+    "type": "object"
+  },
+  "type": [
+    "null",
+    "array"
+  ]
+}

addedOutput schema / properties / grounding

Added value: +{
+  "additionalProperties": false,
+  "description": "Google Search queries the model ran and the pages it used (googleSearch only)",
+  "properties": {
+    "queries": {
+      "items": {
+        "type": "string"
+      },
+      "type": [
+        "null",
+        "array"
+      ]
+    },
+    "sources": {
+      "items": {
+        "additionalProperties": false,
+        "properties": {
+          "title": {
+            "type": "string"
+          },
+          "uri": {
+            "type": "string"
+          }
+        },
+        "required": [
+          "uri"
+        ],
+        "type": "object"
+      },
+      "type": [
+        "null",
+        "array"
+      ]
+    }
+  },
+  "type": [
+    "null",
+    "object"
+  ]
+}

removedOutput schema / properties / mimeType
```
Removed value: -{
-  "type": "string"
-}
```

addedOutput schema / properties / text

Added value: +{
+  "description": "Commentary the model returned alongside the image(s)",
+  "type": "string"
+}

addedOutput schema / properties / warnings

Added value: +{
+  "items": {
+    "type": "string"
+  },
+  "type": [
+    "null",
+    "array"
+  ]
+}

changedOutput schema / required

Previous value: -[
-  "filePath",
-  "model",
-  "mimeType"
-]New value: +[
+  "files",
+  "model",
+  "cost"
+]

First observedApr 15, 2026

TDQS

A4.4/5.0

Behavior4/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Annotations declare non-read-only, open-world, non-idempotent behavior, and the description adds material context beyond that: per-1K pricing for each model (~$0.04 / ~$0.15), the fact that every variation in 'count' is separately billed, and the cost-confirmation threshold that triggers approvedCostUsd. That is genuine cost/behavioral disclosure. It does not cover failure modes or quota limits.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness4/5

Is the description appropriately sized, front-loaded, and free of redundancy?

Three sentences, no filler, with the core capability front-loaded before the model/cost guidance and the downstream-chaining note. It is dense rather than padded, though the final sentence mixes return-value and routing information that the output schema partly covers.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness5/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

For a 9-parameter generative tool with a full output schema and non-trivial cost implications, the description covers capability, cost trade-offs across models, cost-confirmation behavior, and interoperable output URIs. Nothing an agent needs to invoke it correctly is missing, and return-format detail is properly delegated to the output schema.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters4/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema coverage is 100%, so the baseline is 3. The description still adds meaning the schema lacks: the reference-image limit ('up to 14'), the cost ordering of the three model aliases, and the role of the returned URIs relative to edit_image/generate_video. Payments and billing consequences of 'count' are recast in cost terms rather than just a range.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

States a specific verb+resource ('Create images from a text prompt') and immediately scopes it with the reference-image capability. It clearly distinguishes itself from edit_image and generate_video, which it names as downstream consumers of the returned URIs, so the agent can separate generation from editing without opening a schema.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines4/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

Gives concrete model-selection guidance keyed to use case: nb2 as default, pro for 'densest scenes and final renders', nb2-lite for 'bulk drafts'. It also routes the agent to follow-on tools after generation. It stops short of stating when to prefer this tool over edit_image for an iterative change, so it is clear but not fully exclusive.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

Deploy Server