Provider APIVideo and image Q&AVerified 2026-11-05
Gemini 3.1 Flash-Lite
gemini-3.1-flash-lite
Gemini 3.1 Flash-Lite is a video understanding model. Called on Mixpeek through the provider API.
Mixpeek calls the provider's API by name. Provider terms apply.
New opens the Designer in Studio. For an existing namespace, open it in Studio and use its Designer tab.
No demo for Gemini 3.1 Flash-Lite yet
A video, image, slideshow or playground goes here once we record one on our own data.
What it is good and bad at
- Best at
- Low-cost bulk tagging of scenes and frames.
- Worst at
- Weaker than 3.x Flash on long or multi-step video questions.
- Task
- Video and image Q&A
- Input
- Video, Audio, Image, Text, Document
- Output
- Text, Structured data
- Use cases
- Video understanding, Tagging, Captioning, OCR, Brand detection
Templates that use it
No template uses this model yet. Browse templates.
Input and output
Input
- video
- frames
- 1 frame per second by default (static mode); custom fps via the processing config
- tokens_per_frame
- 258 at default media_resolution, 66 at low
- audio
- 32 tokens per second of audio track
- max_length
- about 1 hour at default resolution or 3 hours at low resolution in a 1M-token context
- videos_per_request
- up to 10 (Gemini 2.5 and later)
- formats
- mp4, mpeg, mov, avi, x-flv, mpg, webm, wmv, 3gpp; public YouTube URLs (preview)
- prompts
- free-form instructions; response_schema for JSON output
- max_tokens
- 1M-token input context (Flash family)
Output
- text
- free-form or JSON matching response_schema (tags, descriptions, timestamps MM:SS)
How Mixpeek wires it
- Extractors
- None today
- Feature URIs
- None today
- Vector indexes
- None
- Retriever stages
- llm_enrich, llm_filter, summarize
- Combines with
- scene splits -> Gemini Flash per scene -> tags and descriptions -> E5 text vectors + BM25
- retriever results -> llm_enrich with this model -> structured fields
How it runs
- Status
- Provider API Provider API; accepted by name wherever the platform takes a Gemini model (shared/models/llm/gemini_models.py GEMINI_GENERATION_MODELS). Stable.
- Serve apps
- None
- Runtime
- Google Gen AI SDK (Gemini API / Vertex AI) via shared/models/llm/providers/google.py
- Weights
- Provider API, no remote code
- Parameters
- Not recorded
- Size on disk
- Not recorded
- Preprocessing
- Upload video through the Files API or pass a URI; set fps and media_resolution as needed.
- Postprocessing
- Parse JSON when response_schema is set; meter tokens by modality (shared/billing/model_prices.py).
- Serving
- both
- Hardware
- Provider API
Scale to zero
- Minimum replicas
- Not recorded
- Maximum replicas
- Not recorded
- Idle timeout
- Not recorded
- Memory
- Not recorded
- CPUs
- Not recorded
- GPUs
- Not recorded
- Cold start, node warm
- Not recorded
- Cold start, new node
- Not recorded
Measured cost
Unmeasured
Provider price, not a measurement: input 0.25 (audio 0.50), output 1.50 USD per 1M tokens (server/shared/billing/model_prices.py). Per-video cost is unmeasured.
Our evaluations
Unmeasured
We have not evaluated this model on our own data yet. Publisher results are in the sources below.
Questions
- What is Gemini 3.1 Flash-Lite best at?
- Low-cost bulk tagging of scenes and frames.
- What are the limits of Gemini 3.1 Flash-Lite?
- Weaker than 3.x Flash on long or multi-step video questions.
- Can Gemini 3.1 Flash-Lite be used commercially?
- Yes. Google Gemini API paid-tier terms; outputs usable commercially; paid-tier data is not used to train Google models.
- How does Gemini 3.1 Flash-Lite run on Mixpeek?
- Provider API; accepted by name wherever the platform takes a Gemini model (shared/models/llm/gemini_models.py GEMINI_GENERATION_MODELS). Stable.