Google family
# Gemini 3.1 Pro Advanced multimodal reasoning for high-performance marketing creative. ## Overview Gemini 3.1 Pro represents the latest evolution in Google's flagship generative AI architecture, specifically optimized for complex reasoning, nuanced multimodal understanding, and high-fidelity content generation. Designed for marketing teams and ad agencies, this model excels at synthesizing disparate data points—ranging from raw video footage and high-resolution imagery to intricate brand guidelines—into cohesive, high-converting copy and strategic insights. By leveraging refined multimodal logic, Gemini 3.1 Pro bridges the gap between visual assets and textual output, ensuring that creative outputs remain contextually aligned with your brand's visual identity. Unlike its predecessors, Gemini 3.1 Pro features enhanced performance in long-context window processing and improved adherence to complex system instructions. This makes it an ideal engine for tasks requiring deep analysis of creative briefs, competitor ad analysis, and multi-step content planning. Whether you are generating social media captions that require specific visual context or drafting long-form campaign strategies, this model provides the precision and creative flair necessary for professional-grade marketing workflows. ## Use Cases * **Multimodal Ad Copywriting:** Analyze product images or video storyboards to generate high-converting ad copy that perfectly matches the visual narrative. * **Campaign Strategy Development:** Input brand guidelines and market research data to generate comprehensive, multi-channel campaign strategies. * **Creative Brief Synthesis:** Process complex client briefs and visual assets to produce structured creative direction for design teams. * **Social Media Content Scaling:** Convert single video assets into a suite of platform-specific captions, hashtags, and engagement hooks. * **Competitor Creative Analysis:** Upload competitor ad imagery to identify design trends, messaging patterns, and potential market gaps. * **Brand Voice Alignment:** Use the system prompt to enforce strict brand voice and tone guidelines across all generated marketing collateral. ## Parameters Gemini 3.1 Pro offers a robust suite of configuration options to fine-tune your creative output. You can adjust parameters such as temperature and top_p for creative variance, or utilize the thinking_level and media_resolution settings to balance processing depth with speed. Please refer to the Pixloop parameter interface for the full list of adjustable settings, including seed, penalty controls, and multimodal input handling. ## Tips for Best Results * **Leverage Thinking Levels:** Use the 'HIGH' thinking_level for complex strategy tasks that require deep logical deduction, and 'LOW' for rapid, creative brainstorming sessions. * **Optimize Media Resolution:** When uploading high-fidelity product photography, set media_resolution to 'HIGH' to ensure the model captures fine details, such as text on packaging or specific design motifs. * **Iterative Prompting:** Start with a clear system prompt defining the brand persona, then use the prompt field to provide specific task instructions. If the output is too generic, adjust the frequency_penalty to encourage more unique vocabulary. * **Contextual Anchoring:** Always provide the model with relevant context (e.g., target audience, platform, or campaign goal) within the system_prompt to ensure the generated text is actionable and on-brand. * **Use Stop Sequences:** If you are generating structured formats like JSON or specific ad copy templates, define stop_sequences to prevent the model from generating unnecessary conversational filler. ## About Gemini 3.1 Pro is a premium model provided via Vertex AI, representing the latest iteration in Google's Gemini lineage. Built upon a foundation of massive-scale multimodal training, it is engineered to handle the intersection of text, code, and visual media with unprecedented accuracy. As a preview-tier model, it incorporates the most recent advancements in reasoning architecture, making it a powerful tool for agencies looking to integrate cutting-edge AI into their creative production pipelines.
Model input reference derived from preset schema.
| Name | Type | Required | Default | Description |
|---|---|---|---|---|
| seed | number | No | null | Seed for deterministic output. Leave empty for random. |
| audio | audio | No | null | Input audio to send with the prompt |
| top_p | number | No | 0.95 | Controls diversity via nucleus sampling. |
| images | image | No | [] | Input images to send with the prompt |
| prompt | string | Yes | null | The text prompt to send to the model |
| videos | video | No | [] | Input videos to send with the prompt |
| max_tokens | number | No | 8192 | Maximum number of tokens to generate in the response |
| temperature | number | No | 0.7 | Controls randomness. Lower values are more deterministic. |
| system_prompt | string | No | null | System instructions applied to the model |
| stop_sequences | array | No | null | Sequences that will stop generation (comma separated). |
| thinking_level | select | No | "MEDIUM" | Gemini 3.1 Pro thinking intensity. |
| candidate_count | number | No | 1 | Number of response candidates to generate. |
| media_resolution | select | No | "MEDIUM" | Media processing resolution for multimodal inputs. |
| presence_penalty | number | No | 0 | Penalizes tokens that have already appeared in the response. |
| frequency_penalty | number | No | 0 | Penalizes tokens in proportion to their frequency. |