No public selector verified for this foundation release.
Access
Training announcement with planned Pi deployment. No exact public API selector, release-specific price or downloadable weight artifact is established.
Regions
Current regional availability unverified.
Endpoint pricing has not been collected. Check the provider for applicable rates.
Data & enterprise terms
Retention, training use, regional processing and enterprise terms have not been summarized for this endpoint.
Caching
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.
Inflection research
Not verified
No model-specific cache contract verified.
How it works
Not collected
No model-specific cache contract verified.
Minimum reusable input
Not collected
No model-specific cache contract verified.
Lifetime & refresh
Not collected
No model-specific cache contract verified.
How to enable it
Not collected
No model-specific cache contract verified.
Usage fields
Not collected
No model-specific cache contract verified.
Billing rules
Not collected
No model-specific cache contract verified.
Limits & data boundaries
Not collected
No model-specific cache contract verified.
No verified cache rate for this endpoint. An empty rate is not a zero charge.
What the measurements say
HellaSwag 10-shot
Inflection AI · provider-reported
89%
Score explicitly stated in the release prose. Pre-trained model evaluation; the release cautions that the compared GPT-4 score used a privately held holdout, so the figures are not directly comparable.
Provider-reported pre-trained Inflection-2 accuracy without chain-of-thought prompting; the release specifies 5-shot. Not the subsequently aligned Pi product.
Provider-reported accuracy in Inflection-2 release Table 2. Shot count retained from the column header. The Inflection-1 row is a historical comparison; no chain-of-thought prompting.
Provider-reported accuracy in Inflection-2 release Table 2. Shot count retained from the column header. The Inflection-1 row is a historical comparison; no chain-of-thought prompting.
Provider-reported accuracy in Inflection-2 release Table 2. Shot count retained from the column header. The Inflection-1 row is a historical comparison; no chain-of-thought prompting.
Provider-reported accuracy in Inflection-2 release Table 2. Shot count retained from the column header. The Inflection-1 row is a historical comparison; no chain-of-thought prompting.
Provider-reported accuracy in Inflection-2 release Table 2. Shot count retained from the column header. The Inflection-1 row is a historical comparison; no chain-of-thought prompting.
Provider-reported accuracy in Inflection-2 release Table 2. Shot count retained from the column header. The Inflection-1 row is a historical comparison; no chain-of-thought prompting.
Provider-reported accuracy in Inflection-2 release Table 2. Shot count retained from the column header. The Inflection-1 row is a historical comparison; no chain-of-thought prompting.
Provider-reported pre-trained model results in Inflection-2 release Table 4. The release uses chain-of-thought for GSM8k only. MATH shot count and scoring details need the linked technical memo; no additional settings are inferred.
Provider-reported pre-trained model results in Inflection-2 release Table 4. The release uses chain-of-thought for GSM8k only. MATH shot count and scoring details need the linked technical memo; no additional settings are inferred.
Provider-reported pre-trained model results in Inflection-2 release Table 4. The release uses chain-of-thought for GSM8k only. MATH shot count and scoring details need the linked technical memo; no additional settings are inferred.
Provider-reported pre-trained model results in Inflection-2 release Table 4. The release uses chain-of-thought for GSM8k only. MATH shot count and scoring details need the linked technical memo; no additional settings are inferred.
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
Inferred from the developer’s in-house model or proprietary API adaptation description and restricted access. The reviewed release supplies no weights for this exact checkpoint; this does not classify its dependencies or openly released base models as closed.
