Expand description
Moved. everruns-model-profiles is now part of
everruns-contracts, in the
Everruns ecosystem. This deprecated shim receives
no further updates and leaves the workspace after one migration release.
Replace everruns_model_profiles:: with everruns_contracts::model_profile_data::.
use everruns_contracts::model_profile_data::get_model_profile;
let profile = get_model_profile("anthropic", "claude-sonnet-5").expect("known model");
assert_eq!(profile.family, "claude-sonnet-5");Modules§
Structs§
- Cost
Tier - A pricing tier that activates above a context token threshold. For example, OpenAI charges higher rates for prompts exceeding 200K tokens.
- Model
Cost - Cost information for the model (per million tokens)
- Model
Limits - Token limits for the model
- Model
Modalities - Model modalities for input and output
- Model
Profile Entry - A canonical registry identity and its capability/pricing profile.
- Reasoning
Effort Config - Reasoning effort configuration for a model
- Reasoning
Effort Value - Named reasoning effort value for UI display
- Speed
Config - Speed configuration for a model
- Speed
Value - Named speed value for UI display
- Verbosity
Config - Verbosity configuration for a model
- Verbosity
Value - Named verbosity value for UI display
Enums§
- Modality
- Modality type (text, image, audio, video)
- Model
Vendor - Vendor / brand that authored a model. Independent of the provider type that serves it (the same model may be offered by several providers or gateways). Primarily drives UI iconography.
- Reasoning
Effort - Reasoning effort level for models that support it
- Service
Kind - A typed service a provider driver can offer (see knowledge/foundations/providers.md).
- Speed
- Speed level for models that expose a latency/price service tier.
Wire values map 1:1 to the OpenAI
service_tierrequest parameter:flex(slower, cheaper),default(standard),priority(faster, premium),fast(OpenAI’s current name forpriority; the API accepts both), andultrafast(premium speed tier, model-gated).autois deliberately not offered — omitting the field preserves the provider’s default routing. - Verbosity
- Verbosity level for models that support output-length control.
Wire values map 1:1 to the OpenAI
verbosityrequest parameter:low(terse),medium(balanced, provider default),high(comprehensive). Independent ofReasoningEffort, which tunes the amount of reasoning rather than the length of the final answer.
Constants§
- CLEAR_
AT_ PARAMETER - Model-profile parameter for message expiry.
- MID_
CONVERSATION_ SYSTEM_ PARAMETER - Model-profile parameter for messages-array system roles.
Functions§
- all_
profile_ entries - Enumerate canonical registry entries once each, in registry order.
Profiles are provider-independent, as in
super::get_model_profile_by_key. This is the curated registry, not the entire models.dev catalog; aliases and dated ids do not produce extra entries. Order is not a recommendation ranking. - all_
profiles - Enumerate every canonical profile in registry order, without provider masking.
Use
all_profile_entrieswhen model ids or service kinds are needed. - estimate_
cost_ usd - Estimate the USD cost of a generation from the model’s static price-table
profile, discounting cached-read tokens at the model’s
cache_readrate. ReturnsNonewhen there is no profile or no cost data for the model — callers then have no estimate to record and fall back accordingly. - estimate_
cost_ usd_ for_ speed estimate_cost_usdpriced at the service tier that served the request.- get_
model_ profile - Get a model profile by matching provider_type and model_id. Returns None if the id is not in the registry or is not offered under the given provider type.
- get_
model_ profile_ by_ key - Look up a profile by its stable key (
"{vendor}/{canonical_id}"). - get_
model_ profile_ key - Stable public profile key:
"{vendor}/{canonical_id}"(knowledge/foundations/providers.md). - get_
model_ service_ kind - Which provider service a model belongs to. Unknown models default to
ServiceKind::Chat. - get_
model_ vendor - Get the vendor/brand for a model id, or None if it is not in the registry (or not offered under the given provider type).
- profile_
entries_ for_ provider - Enumerate canonical entries offered under a provider wire id, in registry
order, with the same capability masking as
get_model_profile. Unknown provider ids return an empty list. Registry coverage is not live availability. - profiles_
for_ provider - Enumerate profiles for a provider wire id, applying lookup’s capability masks.
Unknown provider ids return an empty list. Use
profile_entries_for_providerwhen model ids or service kinds are needed. - selected_
profiles - Enumerate the curated selection. Today the registry is itself the selection,
so this is identical to
all_profiles; no independent preset list exists. - selected_
profiles_ for_ provider - Enumerate the curated selection for a provider. Today this is identical to
profiles_for_provider, including its provider capability masking.
Type Aliases§
- Model
Profile Deprecated - Moved to
everruns_contracts::model_profile_data::ModelProfile.