> ## Documentation Index > Fetch the complete documentation index at: https://doc.starrise.tech/llms.txt > Use this file to discover all available pages before exploring further. # Gemini 3 Pro Image Preview (Streaming) > Call Google Gemini 3 Pro Image Preview via Gemini API for streaming image generation. SSE delivers thinking chunks and image chunks in real time. Gemini 3 Pro Image Preview (Streaming) is available through Starrise AI via the native Gemini API, supporting real-time SSE streaming of image generation results. Thinking chunks are pushed first, followed immediately by the final image chunk. ## Key Capabilities * **SSE streaming** — Real-time delivery of thinking chunks and image chunks * **Thinking mode** — Internal reasoning chunks (`thought: true`) streamed before the image * **Text-to-image** — Generate images from text descriptions * **Image editing** — Pass a reference image via `inline_data` combined with text instructions * **Aspect ratio control** — `1:1`, `4:3`, `3:4`, `16:9`, `9:16` * **Resolution control** — `1K` (\~1024px), `2K` (\~2048px), `4K` (\~4096px, by longest side) ## SSE Response Format The streaming endpoint returns newline-delimited SSE data lines, each starting with `data:` followed by a JSON object. There are three chunk types: 1. **Thinking chunk** — Arrives first; `parts[0].thought` is `true` 2. **Image chunk** — Contains `parts[0].inlineData` with `mimeType` and base64 `data` (note: camelCase in streaming responses) 3. **Final usage chunk** — Contains top-level `usageMetadata` with `thoughtsTokenCount` and per-modality token details ``` data: {"candidates":[{"content":{"role":"model","parts":[{"text":"...","thought":true}]}}],"usageMetadata":{"trafficType":"ON_DEMAND"},"modelVersion":"gemini-3-pro-image-preview","createTime":"...","responseId":"..."} data: {"candidates":[{"content":{"role":"model","parts":[{"inlineData":{"mimeType":"image/png","data":""}}]}}],...} data: {"usageMetadata":{"promptTokenCount":8,"candidatesTokenCount":1120,"totalTokenCount":1392,"trafficType":"ON_DEMAND","promptTokensDetails":[{"modality":"TEXT","tokenCount":8}],"candidatesTokensDetails":[{"modality":"IMAGE","tokenCount":1120}],"thoughtsTokenCount":264}} ``` In streaming responses, the image field is `inlineData` (camelCase), while in the request body it is `inline_data` (snake\_case). This is native Gemini API behavior. ## Text-to-Image Example ```bash cURL theme={null} curl "https://ai.alad.com/v1beta/models/gemini-3-pro-image-preview:streamGenerateContent?key=YOUR_API_KEY" \ -H "Content-Type: application/json" \ -d '{ "contents": [ { "role": "user", "parts": [ { "text": "Generate an image of a mountain sunset" } ] } ], "generationConfig": { "responseModalities": ["TEXT", "IMAGE"], "imageConfig": { "aspectRatio": "16:9", "imageSize": "1K" } } }' ``` ```python Python theme={null} import requests, base64, json url = "https://ai.alad.com/v1beta/models/gemini-3-pro-image-preview:streamGenerateContent" params = {"key": "YOUR_API_KEY"} data = { "contents": [ { "role": "user", "parts": [{"text": "Generate an image of a mountain sunset"}] } ], "generationConfig": { "responseModalities": ["TEXT", "IMAGE"], "imageConfig": {"aspectRatio": "16:9", "imageSize": "1K"} } } response = requests.post(url, params=params, json=data, stream=True) for line in response.iter_lines(): if not line: continue decoded = line.decode("utf-8") if not decoded.startswith("data:"): continue chunk = json.loads(decoded[len("data:"):].strip()) candidates = chunk.get("candidates", []) if not candidates: continue parts = candidates[0].get("content", {}).get("parts", []) for part in parts: # Skip thinking chunks if part.get("thought"): continue # Save image chunk if "inlineData" in part: img_bytes = base64.b64decode(part["inlineData"]["data"]) with open("output.png", "wb") as f: f.write(img_bytes) print("Image saved to output.png") ``` ## Image Editing Example (with Reference Image) Pass both a `text` instruction and an `inline_data` reference image in the same `parts` array. ```bash cURL theme={null} # First convert image to base64: # BASE64=$(base64 -i your_photo.jpg) # # Then send the request: curl "https://ai.alad.com/v1beta/models/gemini-3-pro-image-preview:streamGenerateContent?key=YOUR_API_KEY" \ -H "Content-Type: application/json" \ -d '{ "contents": [ { "role": "user", "parts": [ { "text": "This is a photo of me, please add an alpaca beside me" }, { "inline_data": { "mime_type": "image/jpeg", "data": "" } } ] } ], "generationConfig": { "responseModalities": ["TEXT", "IMAGE"], "imageConfig": { "aspectRatio": "1:1", "imageSize": "1K" } } }' ``` ```python Python theme={null} import requests, base64, json # Read and encode the reference image with open("your_photo.jpg", "rb") as f: image_b64 = base64.b64encode(f.read()).decode("utf-8") url = "https://ai.alad.com/v1beta/models/gemini-3-pro-image-preview:streamGenerateContent" params = {"key": "YOUR_API_KEY"} data = { "contents": [ { "role": "user", "parts": [ { "text": "This is a photo of me, please add an alpaca beside me" }, { "inline_data": { "mime_type": "image/jpeg", "data": image_b64 } } ] } ], "generationConfig": { "responseModalities": ["TEXT", "IMAGE"], "imageConfig": {"aspectRatio": "1:1", "imageSize": "1K"} } } response = requests.post(url, params=params, json=data, stream=True) for line in response.iter_lines(): if not line: continue decoded = line.decode("utf-8") if not decoded.startswith("data:"): continue chunk = json.loads(decoded[len("data:"):].strip()) candidates = chunk.get("candidates", []) if not candidates: continue parts = candidates[0].get("content", {}).get("parts", []) for part in parts: if part.get("thought"): continue if "inlineData" in part: img_bytes = base64.b64decode(part["inlineData"]["data"]) with open("output.png", "wb") as f: f.write(img_bytes) print("Image saved to output.png") ``` ## Parameters | Parameter | Type | Required | Description | | ------------------------------------------ | ------ | -------- | ---------------------------------------------------------------------------------------- | | `key` | string | Yes | API key (query parameter) | | `alt` | string | No | Set to `sse` to explicitly enable SSE mode (optional, streaming is the default behavior) | | `contents[].parts[].text` | string | Yes | Text prompt or instruction | | `contents[].parts[].inline_data.mime_type` | string | No | Reference image type: `image/jpeg`, `image/png`, `image/webp` | | `contents[].parts[].inline_data.data` | string | No | Base64-encoded reference image data | | `generationConfig.responseModalities` | array | Yes | `["IMAGE"]` or `["TEXT", "IMAGE"]` | | `generationConfig.imageConfig.aspectRatio` | string | No | `1:1` / `4:3` / `3:4` / `16:9` / `9:16` | | `generationConfig.imageConfig.imageSize` | string | No | `1K` / `2K` / `4K` (default `1K`) | View the interactive API Playground for Gemini 3 Pro Image Preview (Streaming).