NEWVectors or files. Pick a path.Start →
    Provider APIClassificationVerified 2026-11-05

    TypeSafe Jev

    jev-latest

    TypeSafe Jev is a text and structured classification model (TypeSafe AI API terms). Called on Mixpeek through the provider API.

    Mixpeek calls the provider's API by name. Provider terms apply.

    Use in a new namespaceAdd to an existing namespace Record as JSON

    New opens the Designer in Studio. For an existing namespace, open it in Studio and use its Designer tab.

    No demo for TypeSafe Jev 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
    Typed decisions with calibrated probabilities over a JSON state: yes/no relevance for rerank, label probabilities for classify, closed-set schema fields for text_extractor; the platform's DECISION_MODEL_DEFAULT.
    Worst at
    Free-text generation and open-ended fields (closed-set answers only); no image, video or audio input; 32k-token state limit (MAX_STATE_CHARS 100,000).
    Task
    Classification
    Input
    Text, Structured data
    Output
    Labels, Scores, Structured data
    Use cases
    Classification, Reranking, Moderation

    Templates that use it

    No template uses this model yet. Browse templates.

    Input and output

    Input

    state
    JSON object or text, up to about 100,000 characters
    questions
    typed questions (choice, yes/no, schema fields)

    Output

    structured
    answers with per-option probabilities; rerank -> {scores}, classify -> {labels: [{label, confidence}]}

    How Mixpeek wires it

    Extractors
    • text_extractor@v1 llm_provider=typesafe or llm_model starting with jev (response_shape set)
    Feature URIs
    None today
    Vector indexes
    None
    Retriever stages
    rerank, llm_enrich, classify
    Combines with
    • feature_search -> rerank inference_name=typesafe__jev (or mixpeek__decision, default jev-latest)

    How it runs

    Status
    Provider API TypeSafe System One HTTP API. typesafe__jev registers the name; on the planes mixpeek__api_inference hosts it as an API target.
    Serve apps
    typesafe__jev
    Runtime
    httpx client to https://api.typesafe.ai/v1/systemone (clients.providers.typesafe.client)
    Weights
    Provider API, no remote code
    Parameters
    Not recorded
    Size on disk
    Not recorded
    Preprocessing
    One question per (query, document) pair for rerank; labels become a choice question for classify; schema fields become typed questions.
    Postprocessing
    Yes-probability as the rerank score; label probabilities sorted highest first.
    Serving
    realtime
    Hardware
    Provider API

    Scale to zero

    Minimum replicas
    0
    Maximum replicas
    4
    Idle timeout
    600 s
    Memory
    Not recorded
    CPUs
    0.1
    GPUs
    0
    Cold start, node warm
    Not recorded
    Cold start, new node
    Not recorded

    Measured cost

    Unmeasured

    Price per client.py is 0.042 USD per million input tokens, output free; no committed throughput or latency measurement.

    Our evaluations

    Unmeasured

    We have not evaluated this model on our own data yet. Publisher results are in the sources below.

    Questions

    What is TypeSafe Jev best at?
    Typed decisions with calibrated probabilities over a JSON state: yes/no relevance for rerank, label probabilities for classify, closed-set schema fields for text_extractor; the platform's DECISION_MODEL_DEFAULT.
    What are the limits of TypeSafe Jev?
    Free-text generation and open-ended fields (closed-set answers only); no image, video or audio input; 32k-token state limit (MAX_STATE_CHARS 100,000).
    Can TypeSafe Jev be used commercially?
    Yes. Paid TypeSafe API (input tokens only, 0.042 USD per million per client.py); used under TypeSafe's terms.
    How does TypeSafe Jev run on Mixpeek?
    TypeSafe System One HTTP API. typesafe__jev registers the name; on the planes mixpeek__api_inference hosts it as an API target.