Runs as a custom pluginEmbeddingVerified 2026-11-05
OpenAI CLIP ViT-L/14
openai/clip-vit-large-patch14
OpenAI CLIP ViT-L/14 is a image and text embedding model (MIT). Can run on Mixpeek as a custom plugin.
Mixpeek does not run it today. It can run as a custom plugin: the weights load without remote code.
No demo for OpenAI CLIP ViT-L/14 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
- General zero-shot labels and text-to-image search; the reference baseline most papers report against.
- Worst at
- English only; weaker than SigLIP and LAION OpenCLIP at equal size on retrieval; reads text in images poorly.
- Task
- Embedding
- Input
- Image, Text
- Output
- Vector
- Use cases
- Visual search, Product search
Templates that use it
No template uses this model yet. Browse templates.
Input and output
Input
- image
- resolution
- 224x224
- color
- RGB
- normalize
- CLIP mean (0.4815, 0.4578, 0.4082), std (0.2686, 0.2613, 0.2758)
- text
- max_tokens
- 77
- prompts
- 'a photo of a {label}.'; BPE tokenizer, 77 tokens max
Output
- vector
- dims
- 768
- distance
- cosine
- type
- dense
- note
- image and text vectors share one space; L2-normalize before cosine
How Mixpeek wires it
- Extractors
- None today
- Feature URIs
- None today
- Vector indexes
- None
- Retriever stages
- feature_search, mmr
- Combines with
- video frames (VideoFrameExpander) -> image vectors -> feature_search with a text query
- detector boxes (RT-DETR, D-FINE) -> crops -> image vectors per object
- image vectors + E5 text vectors on captions -> feature_search fusion (dbsf)
How it runs
- Status
- Runs as a custom plugin Not served. Would run as a custom extractor plugin (engine/extractors/<name>/v1/ with realtime.py for query embedding), loading safetensors with trust_remote_code false.
- Serve apps
- None
- Runtime
- transformers CLIPModel + CLIPProcessor
- Weights
- safetensors, no remote code
- Parameters
- 428 M
- Size on disk
- 1,711 MB
- Preprocessing
- Resize shortest side and center-crop to 224x224, normalize with CLIP mean/std; text to 77 tokens.
- Postprocessing
- L2-normalize image and text embeddings; logits = 100 * cosine for zero-shot.
- Serving
- both, batch 32
- Hardware
- Runs on CPU (slower without a GPU)
Scale to zero
- Minimum replicas
- 0
- Maximum replicas
- 1
- Idle timeout
- 600 s
- Memory
- 3 GB
- CPUs
- 1
- GPUs
- 0
- Cold start, node warm
- Not recorded
- Cold start, new node
- Not recorded
Measured cost
Unmeasured
No committed measurement; plan row 3 measures it in the standalone image.
Our evaluations
Unmeasured
We have not evaluated this model on our own data yet. Publisher results are in the sources below.
Questions
- What is OpenAI CLIP ViT-L/14 best at?
- General zero-shot labels and text-to-image search; the reference baseline most papers report against.
- What are the limits of OpenAI CLIP ViT-L/14?
- English only; weaker than SigLIP and LAION OpenCLIP at equal size on retrieval; reads text in images poorly.
- Can OpenAI CLIP ViT-L/14 be used commercially?
- Yes. MIT license on code and weights; the OpenAI model card says deployed use cases are out of scope and calls the release research-oriented, so treat production use as a judgment call.
- How does OpenAI CLIP ViT-L/14 run on Mixpeek?
- Not served. Would run as a custom extractor plugin (engine/extractors/<name>/v1/ with realtime.py for query embedding), loading safetensors with trust_remote_code false.
- What does OpenAI CLIP ViT-L/14 output?
- A 768-dimensional dense vector, compared by cosine similarity.