Provider-documented facts. “Not collected” means we have not verified that field yet, not that the feature is unsupported.
Context / input limit
Not collected
An exact-version public contract was not established. Agent capabilities, underlying base-model limits and provider-side RL are not standalone model API guarantees.
Maximum input tokens
Not collected
An exact-version public contract was not established. Agent capabilities, underlying base-model limits and provider-side RL are not standalone model API guarantees.
Maximum output
Not collected
An exact-version public contract was not established. Agent capabilities, underlying base-model limits and provider-side RL are not standalone model API guarantees.
Parameters
Not collected
An exact-version public contract was not established. Agent capabilities, underlying base-model limits and provider-side RL are not standalone model API guarantees.
Used within the provider agent harness; this does not establish arbitrary public function-schema support.
Structured output
Not collected
An exact-version public contract was not established. Agent capabilities, underlying base-model limits and provider-side RL are not standalone model API guarantees.
Streaming
Not collected
An exact-version public contract was not established. Agent capabilities, underlying base-model limits and provider-side RL are not standalone model API guarantees.
Batch processing
Not collected
An exact-version public contract was not established. Agent capabilities, underlying base-model limits and provider-side RL are not standalone model API guarantees.
Fine-tuning
Not collected
An exact-version public contract was not established. Agent capabilities, underlying base-model limits and provider-side RL are not standalone model API guarantees.
Context caching
Not collected
An exact-version public contract was not established. Agent capabilities, underlying base-model limits and provider-side RL are not standalone model API guarantees.
Knowledge cutoff
Not collected
An exact-version public contract was not established. Agent capabilities, underlying base-model limits and provider-side RL are not standalone model API guarantees.
No current public checkpoint selector established.
Access
Historical product release; current selection, entitlement and numeric price are unverified.
Regions
Model-specific processing region not established; dedicated enterprise deployment options depend on the agreement.
Endpoint pricing has not been collected. Check the provider for applicable rates.
Data & enterprise terms
Self-serve data may be used for model training by default. Paid customers can opt out; Teams requires an admin. Enterprise data is not used for training without prior written consent. Model-provider zero retention after opt-out is not zero retention of all Devin account/session data; customer data and feedback have separate retention rules.
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.
Windsurf · historical release
Not verified
Exact-version cache support and billing contract not established. Internal KV reuse and retrieved code are not proof of discounted prompt caching.
How it works
Not collected
Exact-version cache support and billing contract not established. Internal KV reuse and retrieved code are not proof of discounted prompt caching.
Minimum reusable input
Not collected
Exact-version cache support and billing contract not established. Internal KV reuse and retrieved code are not proof of discounted prompt caching.
Lifetime & refresh
Not collected
Exact-version cache support and billing contract not established. Internal KV reuse and retrieved code are not proof of discounted prompt caching.
How to enable it
Not collected
Exact-version cache support and billing contract not established. Internal KV reuse and retrieved code are not proof of discounted prompt caching.
Usage fields
Not collected
Exact-version cache support and billing contract not established. Internal KV reuse and retrieved code are not proof of discounted prompt caching.
Billing rules
Not collected
Exact-version cache support and billing contract not established. Internal KV reuse and retrieved code are not proof of discounted prompt caching.
Limits & data boundaries
Not collected
Exact-version cache support and billing contract not established. Internal KV reuse and retrieved code are not proof of discounted prompt caching.
No verified cache rate for this endpoint. An empty rate is not a zero charge.
Caching reviewed 15 Oct 2026
What the measurements say
No benchmark observations have been added for this model. We do not turn a vendor description into a performance score.
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 a distinct in-house trained model supplied through Windsurf/Devin. Private-weight classification is inferred from the product/research distribution and proprietary service terms; the reviewed sources do not publish an exact checkpoint weight license. A public pretrained base is not a release of the SWE derivative. This inference must be revisited if derivative weights are released.
