Answer visual questions
Produce a text answer to a question grounded in one or more supplied images.
Desired outcome
A question-specific textual interpretation of visual evidence.
Explore from the outcome
Where this function can lead
These documented connections are starting points for selection, not a ranked list.
Model releases
Identifiable models documented for this function.
- Claude Fable 5.1
Demanding reasoning and long-horizon agentic work with text responses.
Link evidence: Claude models overview
- Claude Haiku 4.5
Fast text and image understanding with text responses for routine tasks.
Link evidence: Claude models overview
- Claude Opus 5.5
Long-running agentic coding and knowledge work with text responses.
Link evidence: Claude models overview
- Claude Sonnet 5.5
General reasoning, coding, and knowledge tasks with an emphasis on speed and capability.
Link evidence: Claude models overview
- Gemini 3.1 Pro Preview
Multimodal reasoning and software engineering with text output through the Gemini API.
Link evidence: Gemini 3.1 Pro Preview model page
- Gemini 3.8 Flash
Fast multimodal text response generation for coding and agentic tasks.
Link evidence: Gemini 3.8 Flash model page
- Gemma 3n E2B IT LiteRT preview
Produces text from text or images with a Google AI Edge preview checkpoint prepared for Android deployment.
Link evidence: Google Gemma 3n E2B LiteRT preview model card, Google AI Edge Gallery model allowlist
- GPT-6 Astra
Reasons over text and images and generates text through OpenAI's hosted API under the exact gpt-6-astra model ID.
Link evidence: OpenAI images and vision guide, GPT-6 Astra official model page
- GPT-6 Luna
Focused, high-volume reasoning and text generation through OpenAI API.
Link evidence: GPT-6 Luna official model page
- GPT-6.1 Sol
General reasoning, coding, and document work through a current OpenAI API model.
Link evidence: GPT-6.1 Sol official model page
- Llama 4 Maverick 17B-128E
Generate text from text and image prompts using Meta’s Maverick multimodal release.
Link evidence: Llama 4 Maverick model card
- Llama 4 Scout 17B-16E
Generate text from text and image prompts using Meta’s Scout multimodal release.
Link evidence: Llama 4 Scout model card
- Mistral Large 3
General-purpose multimodal understanding and text generation with Mistral Large 3.
Link evidence: Mistral Large 3 model card
- Qwen3.5-397B-A17B
Vision-language reasoning and text generation with Qwen’s 397B-A17B release.
Link evidence: Qwen3.5-397B-A17B model card
- Qwen3.5-9B
Vision-language reasoning and text generation with Qwen’s 9B release.
Link evidence: Qwen3.5-9B model card
- SmolVLM Instruct
Answers visual questions and describes images in text with a compact vision-language model.
Link evidence: SmolVLM-Instruct publisher model card
Typical inputs
- One or more images and a text question
Typical outputs
- Text answer grounded in the supplied images
Sources
Last reviewed 2026-09-30