No standalone model selector or immutable inference version is published in the reviewed sources.
Access
Requires Glean Protect+ entitlement and applicable feature configuration; no standalone detector tariff or native model API verified.
Regions
Deployment-dependent single-tenant infrastructure. The subprocessor list allows customer-selected cloud location; third-party processing and support locations can differ.
Endpoint pricing has not been collected. Check the provider for applicable rates.
Data & enterprise terms
Current June 2026 terms govern service access and customer-data processing; separate signed agreements may supersede them. The AI Addendum prohibits training on Customer Data for covered AI Models and promises zero-day retention for Glean-key use, with law and trust/safety exceptions. Customer-key access is the customer’s responsibility. The July Limited Retention Addendum preserves no-training restrictions while allowing specified retention; current docs identify Claude Fable as an opt-in limited-retention model. Native search-encoder training scope, request retention periods and customer-specific terms remain unresolved.
Caching can reuse earlier input processing. Availability, lifetime and billing depend on the endpoint. A cache hit still generates a new response.
01 / INITIAL INPUTProcess & retain
The system processes eligible input and retains reusable state.
02 / LATER REQUESTCheck available state
Reuse depends on matching input, session state and the cache policy.
03 / CACHE HITReuse & generate
Reusing state can reduce repeated work. Any charges follow the billing rules below.
Glean enterprise platform
Not verified
No model-specific inference-cache contract verified. LLM event cache counters, search indexes, application memory and retention are not proof of native cache support for this model.
How it works
Not collected
No model-specific inference-cache contract verified. LLM event cache counters, search indexes, application memory and retention are not proof of native cache support for this model.
Minimum reusable input
Not collected
No model-specific inference-cache contract verified. LLM event cache counters, search indexes, application memory and retention are not proof of native cache support for this model.
Lifetime & refresh
Not collected
No model-specific inference-cache contract verified. LLM event cache counters, search indexes, application memory and retention are not proof of native cache support for this model.
How to enable it
Not collected
No model-specific inference-cache contract verified. LLM event cache counters, search indexes, application memory and retention are not proof of native cache support for this model.
Usage fields
Not collected
No model-specific inference-cache contract verified. LLM event cache counters, search indexes, application memory and retention are not proof of native cache support for this model.
Billing rules
Not collected
No model-specific inference-cache contract verified. LLM event cache counters, search indexes, application memory and retention are not proof of native cache support for this model.
Limits & data boundaries
Not collected
No model-specific inference-cache contract verified. LLM event cache counters, search indexes, application memory and retention are not proof of native cache support for this model.
No verified cache rate for this endpoint. An empty rate is not a zero charge.
Caching reviewed 16 Oct 2026
What the measurements say
Direct prompt-injection accuracy
Glean · provider-reported
97.8%
September 2025 provider evaluation across its described task-specific datasets and verification pipeline. Not independently reproduced; not a universal production guarantee.
September 2025 provider evaluation across its described task-specific datasets and verification pipeline. Not independently reproduced; not a universal production guarantee.
September 2025 provider evaluation across its described task-specific datasets and verification pipeline. Not independently reproduced; not a universal production guarantee.
September 2025 provider evaluation across its described task-specific datasets and verification pipeline. Not independently reproduced; not a universal production guarantee.
Latency and throughput depend on the host, request size, reasoning settings and load. There is no measured speed or global usage figure in this record.
The record behind the model
Why it is included
Provider describes internally trained or adapted models used within its commercial service. Underlying technology is reserved under current terms; no public derivative checkpoint is established. This record represents a model family, not a count of unnamed versions or customer copies.
