diff --git a/litellm-rust/crates/cost/src/completion_input.rs b/litellm-rust/crates/cost/src/completion_input.rs index a0e704c1f58..24c329541c1 100644 --- a/litellm-rust/crates/cost/src/completion_input.rs +++ b/litellm-rust/crates/cost/src/completion_input.rs @@ -6,6 +6,7 @@ use crate::error::CostError; use crate::model_selection::{ ModelSelectionRequest, get_response_model, select_model_name_for_cost_calc, }; +use crate::provider::LlmProviders; use crate::responses_usage::ChatUsage; use crate::usage_dispatch::get_usage_object; @@ -111,7 +112,7 @@ pub fn prepare_completion_input( infer_call_type(request.call_type, request.response_kind).unwrap_or("completion"); let model = if matches!(call_type, "image_generation" | "aimage_generation") && request.model_selection.model == Some("") - && request.model_selection.provider == Some("azure") + && LlmProviders::AZURE.matches(request.model_selection.provider) { Some("dall-e-2") } else { diff --git a/litellm-rust/crates/cost/src/completion_response.rs b/litellm-rust/crates/cost/src/completion_response.rs index 632107f95c1..6e3698cd6d2 100644 --- a/litellm-rust/crates/cost/src/completion_response.rs +++ b/litellm-rust/crates/cost/src/completion_response.rs @@ -24,6 +24,7 @@ use crate::image_cost_router::{ }; use crate::mcp_cost::calculate_mcp_tool_call_cost; use crate::per_second::{DEFAULT_REPLICATE_GPU_PRICE_PER_SECOND, get_replicate_completion_pricing}; +use crate::provider::LlmProviders; use crate::realtime_cost::{ collect_and_combine_usage_from_realtime_stream_results, combine_usage_objects, event_usage, partition_results_by_service_tier, @@ -410,7 +411,7 @@ fn unregistered_replicate_cost( provider: Option<&str>, ) -> Option<(String, f64)> { let model = prepared.model_candidates.iter().flatten().next()?; - if !(provider == Some("replicate") || model.contains("replicate")) + if !(LlmProviders::REPLICATE.matches(provider) || model.contains("replicate")) || catalog.contains_exact_model(model) { return None; @@ -768,7 +769,7 @@ pub fn completion_cost_from_response( .filter(|model| !model.is_empty()) .or_else(|| hidden.get("litellm_model_name").and_then(Value::as_str)) }); - let request_model = if provider == Some("azure_ai") { + let request_model = if LlmProviders::AZURE_AI.matches(provider) { hidden_model .filter(|model| is_azure_model_router(model)) .or(request.request_model) @@ -790,7 +791,7 @@ pub fn completion_cost_from_response( ), |model| { let token_request_model = - if provider == Some("azure_ai") && !is_azure_model_router(model) { + if LlmProviders::AZURE_AI.matches(provider) && !is_azure_model_router(model) { None } else { request_model @@ -819,10 +820,11 @@ pub fn completion_cost_from_response( cost_per_token_for_call(catalog, cost_request, call) }, )?; - let router_fee = (!is_search && provider == Some("azure_ai") && !is_azure_model_router(&model)) - .then(|| azure_ai_router_fee(catalog, &model, request_model, usage.prompt_tokens)) - .and_then(Result::ok) - .flatten(); + let router_fee = + (!is_search && LlmProviders::AZURE_AI.matches(provider) && !is_azure_model_router(&model)) + .then(|| azure_ai_router_fee(catalog, &model, request_model, usage.prompt_tokens)) + .and_then(Result::ok) + .flatten(); let supplied_additional_costs = if is_search { &[] } else { diff --git a/litellm-rust/crates/cost/src/cost_calculator.rs b/litellm-rust/crates/cost/src/cost_calculator.rs index a5f6eb63ae2..fa73e47eb9b 100644 --- a/litellm-rust/crates/cost/src/cost_calculator.rs +++ b/litellm-rust/crates/cost/src/cost_calculator.rs @@ -241,7 +241,7 @@ pub fn cost_per_token( None => catalog.select_model_key(request.model, request.provider, request.region), } .or_else(|| { - (request.provider == Some("fireworks_ai")) + (provider == Some(LlmProviders::FIREWORKS_AI)) .then(|| get_base_model_for_pricing(request.model, FireworksThresholds::default())) .and_then(|category| { catalog.select_model_key(category, request.provider, request.region) @@ -354,8 +354,13 @@ pub fn speech_cost( let model_info = catalog .entry(request.model, request.provider, request.region) .ok_or(CostError::ModelNotFound)?; - if matches!(request.provider, Some("vertex_ai" | "vertex_ai_beta")) - && let Some(cost) = lyria_generation_cost(model_info) + let provider = request + .provider + .and_then(|provider| provider.parse::().ok()); + if matches!( + provider, + Some(LlmProviders::VERTEX_AI | LlmProviders::VERTEX_AI_BETA) + ) && let Some(cost) = lyria_generation_cost(model_info) { return Ok((0.0, cost)); } diff --git a/litellm-rust/crates/cost/src/provider.rs b/litellm-rust/crates/cost/src/provider.rs index c400c1daf3c..eea3a0c24d9 100644 --- a/litellm-rust/crates/cost/src/provider.rs +++ b/litellm-rust/crates/cost/src/provider.rs @@ -340,4 +340,8 @@ impl LlmProviders { pub fn as_str(self) -> &'static str { self.into() } + + pub fn matches(self, name: Option<&str>) -> bool { + name == Some(self.as_str()) + } } diff --git a/litellm-rust/crates/cost/src/provider_cache.rs b/litellm-rust/crates/cost/src/provider_cache.rs index 5722349ae43..65034f48c4c 100644 --- a/litellm-rust/crates/cost/src/provider_cache.rs +++ b/litellm-rust/crates/cost/src/provider_cache.rs @@ -1,5 +1,7 @@ use serde_json::{Number, Value}; +use crate::provider::LlmProviders; + fn rate(value: Option<&Value>) -> Option { match value? { Value::Number(value) => value.as_f64(), @@ -61,7 +63,7 @@ pub fn with_default_cache_read_rate(model_info: &Value) -> Value { } pub fn apply_provider_cache_read_default(model_info: &Value, provider: Option<&str>) -> Value { - if provider == Some("fireworks_ai") { + if LlmProviders::FIREWORKS_AI.matches(provider) { with_default_cache_read_rate(model_info) } else { model_info.clone() diff --git a/litellm-rust/crates/cost/src/tool_call_cost_tracking.rs b/litellm-rust/crates/cost/src/tool_call_cost_tracking.rs index fc2dbeb17f1..199ca59692a 100644 --- a/litellm-rust/crates/cost/src/tool_call_cost_tracking.rs +++ b/litellm-rust/crates/cost/src/tool_call_cost_tracking.rs @@ -1,5 +1,7 @@ use serde_json::Value; +use crate::provider::LlmProviders; + #[derive(Clone, Copy, Debug, PartialEq)] pub struct DefaultToolRates { pub file_search_per_call: f64, @@ -212,7 +214,7 @@ pub fn get_cost_for_file_search( return 0.0; } let storage = storage_gb.unwrap_or(0.0) * days.unwrap_or(0.0); - if provider == Some("azure") + if LlmProviders::AZURE.matches(provider) && let Some(rate) = model_info.and_then(|info| number(info.get("file_search_cost_per_gb_per_day"))) { @@ -223,7 +225,7 @@ pub fn get_cost_for_file_search( { return rate; } - if provider == Some("azure") { + if LlmProviders::AZURE.matches(provider) { return storage * defaults.azure_file_search_per_gb_day; } defaults.file_search_per_call @@ -243,7 +245,9 @@ pub fn get_cost_for_vector_store( let rate = model_info.and_then(|info| number(info.get("vector_store_cost_per_gb_per_day"))); match rate { Some(rate) => storage * rate, - None if provider == Some("azure") => storage * defaults.azure_vector_store_per_gb_day, + None if LlmProviders::AZURE.matches(provider) => { + storage * defaults.azure_vector_store_per_gb_day + } None => 0.0, } } @@ -255,7 +259,7 @@ pub fn get_cost_for_computer_use( model_info: Option<&Value>, defaults: DefaultToolRates, ) -> f64 { - if provider != Some("azure") { + if !LlmProviders::AZURE.matches(provider) { return 0.0; } let input_rate = model_info diff --git a/litellm-rust/crates/cost/src/tool_cost_dispatch.rs b/litellm-rust/crates/cost/src/tool_cost_dispatch.rs index 73c99b64dd0..0d44eee0b19 100644 --- a/litellm-rust/crates/cost/src/tool_cost_dispatch.rs +++ b/litellm-rust/crates/cost/src/tool_cost_dispatch.rs @@ -6,6 +6,7 @@ use crate::anthropic_cost::{ }; use crate::gemini_cost::{cost_per_google_maps_grounding_request, cost_per_web_search_request}; use crate::groq_cost::cost_per_web_search_request as groq_web_search_cost; +use crate::provider::LlmProviders; use crate::responses_usage::ChatUsage; use crate::tool_call_cost_tracking::{ DefaultToolRates, ResponseKind, count_web_search_calls, extract_file_search_params, @@ -115,7 +116,7 @@ fn provider_web_search_cost( } fn maps_cost(request: BuiltInToolCostRequest<'_>, model_info: Option<&Value>) -> f64 { - let supports_maps = matches!(request.provider, Some("gemini")) + let supports_maps = LlmProviders::GEMINI.matches(request.provider) || request .provider .is_some_and(|provider| provider.starts_with("vertex_ai")); @@ -149,7 +150,7 @@ fn file_search_cost(request: BuiltInToolCostRequest<'_>, model_info: Option<&Val } fn azure_assistant_cost(request: BuiltInToolCostRequest<'_>, model_info: Option<&Value>) -> f64 { - if request.provider != Some("azure") { + if !LlmProviders::AZURE.matches(request.provider) { return 0.0; } let vector = request