modelcatalog

package
v1.2.37 Latest Latest
Warning

This package is not in the latest version of its module.

 Go to latest Published: Apr 10, 2026 License: Apache-2.0 Imports: 17 Imported by: 9

Documentation

Overview

Package modelcatalog provides a pricing manager for the framework.

Index

Constants

View Source 
const (
	DefaultPricingSyncInterval    = 24 * time.Hour
	MinimumPricingSyncIntervalSec = int64(3600)

	ConfigLastPricingSyncKey      = "LastModelPricingSync"
	ConfigLastParamsSyncKey       = "LastModelParametersSync"
	DefaultPricingURL             = "https://getbifrost.ai/datasheet"
	DefaultModelParametersURL     = "https://getbifrost.ai/datasheet/model-parameters"
	DefaultPricingTimeout         = 45 * time.Second
	DefaultModelParametersTimeout = 45 * time.Second
)
View Source 
const (
	TokenTierAbove272K = 272000
	TokenTierAbove200K = 200000
	TokenTierAbove128K = 128000
)

Default sync interval and config key

Variables

This section is empty.

Functions

This section is empty.

Types

type Config 

type Config struct {
	PricingURL          *string `json:"pricing_url,omitempty"`
	PricingSyncInterval *int64  `json:"pricing_sync_interval,omitempty"` // seconds
}

Config is the model pricing configuration.

type ModelCatalog 

type ModelCatalog struct {
	// contains filtered or unexported fields
}

func Init 

func Init(ctx context.Context, config *Config, configStore configstore.ConfigStore, shouldSyncPricingFunc ShouldSyncPricingFunc, logger schemas.Logger) (*ModelCatalog, error)

Init initializes the model catalog

func NewTestCatalog added in v1.2.19 

func NewTestCatalog(baseModelIndex map[string]string) *ModelCatalog

NewTestCatalog creates a minimal ModelCatalog for testing purposes. It does not start background sync workers or connect to external services.

func (*ModelCatalog) CalculateCost 

func (mc *ModelCatalog) CalculateCost(result *schemas.BifrostResponse) float64

CalculateCost calculates the cost of a Bifrost response. It handles all request types, cache debug billing, and tiered pricing.

func (*ModelCatalog) Cleanup 

func (mc *ModelCatalog) Cleanup() error

Cleanup cleans up the model catalog

func (*ModelCatalog) DeleteModelDataForProvider added in v1.1.21 

func (mc *ModelCatalog) DeleteModelDataForProvider(provider schemas.ModelProvider)

DeleteModelDataForProvider deletes all model data from the pool for a given provider

func (*ModelCatalog) DeleteProviderPricingOverrides added in v1.2.22 

func (mc *ModelCatalog) DeleteProviderPricingOverrides(provider schemas.ModelProvider)

func (*ModelCatalog) ForceReloadPricing added in v1.1.48 

func (mc *ModelCatalog) ForceReloadPricing(ctx context.Context) error

func (*ModelCatalog) GetBaseModelName added in v1.2.19 

func (mc *ModelCatalog) GetBaseModelName(model string) string

GetBaseModelName returns the canonical base model name for a given model string. It uses the pre-computed base_model from the pricing catalog when available, falling back to algorithmic date/version stripping for models not in the catalog.

Examples: 

mc.GetBaseModelName("gpt-4o")                    // Returns: "gpt-4o"
mc.GetBaseModelName("openai/gpt-4o")             // Returns: "gpt-4o"
mc.GetBaseModelName("gpt-4o-2024-08-06")         // Returns: "gpt-4o" (algorithmic fallback)

func (*ModelCatalog) GetDistinctBaseModelNames added in v1.2.19 

func (mc *ModelCatalog) GetDistinctBaseModelNames() []string

GetDistinctBaseModelNames returns all unique base model names from the catalog (thread-safe). This is used for governance model selection when no specific provider is chosen.

func (*ModelCatalog) GetModelCapabilityEntryForModel added in v1.2.35 

func (mc *ModelCatalog) GetModelCapabilityEntryForModel(model string, provider schemas.ModelProvider) *PricingEntry

GetModelCapabilityEntryForModel returns capability metadata for a model/provider pair. It prefers chat, then responses, then text-completion entries; if none exist, it falls back to the lexicographically first available mode for deterministic behavior.

func (*ModelCatalog) GetModelsForProvider 

func (mc *ModelCatalog) GetModelsForProvider(provider schemas.ModelProvider) []string

GetModelsForProvider returns all available models for a given provider (thread-safe)

func (*ModelCatalog) GetPricingEntryForModel added in v1.1.28 

func (mc *ModelCatalog) GetPricingEntryForModel(model string, provider schemas.ModelProvider) *PricingEntry

GetPricingEntryForModel returns the pricing data

func (*ModelCatalog) GetProvidersForModel 

func (mc *ModelCatalog) GetProvidersForModel(model string) []schemas.ModelProvider

GetProvidersForModel returns all providers for a given model (thread-safe)

func (*ModelCatalog) GetUnfilteredModelsForProvider added in v1.2.22 

func (mc *ModelCatalog) GetUnfilteredModelsForProvider(provider schemas.ModelProvider) []string

GetUnfilteredModelsForProvider returns all available models for a given provider (thread-safe)

func (*ModelCatalog) IsModelAllowedForProvider added in v1.2.9 

func (mc *ModelCatalog) IsModelAllowedForProvider(provider schemas.ModelProvider, model string, allowedModels []string) bool

IsModelAllowedForProvider checks if a model is allowed for a specific provider based on the allowed models list and catalog data. It handles all cross-provider logic including provider-prefixed models and special routing rules.

Parameters:

  • provider: The provider to check against
  • model: The model name (without provider prefix, e.g., "gpt-4o" or "claude-3-5-sonnet")
  • allowedModels: List of allowed model names (can be empty, can include provider prefixes)

Behavior:

  • If allowedModels is empty: Uses model catalog to check if provider supports the model (delegates to GetProvidersForModel which handles all cross-provider logic)
  • If allowedModels is not empty: Checks if model matches any entry in the list Provider-specific validation:
  • Direct matches: "gpt-4o" in allowedModels for any provider
  • Prefixed matches: Only if the prefixed model exists in provider's catalog (e.g., "openai/gpt-4o" in allowedModels only matches if openrouter's catalog contains "openai/gpt-4o" AND the model part matches the request)

Returns:

  • bool: true if the model is allowed for the provider, false otherwise

Examples: 

// Empty allowedModels - uses catalog
mc.IsModelAllowedForProvider("openrouter", "claude-3-5-sonnet", []string{})
// Returns: true (catalog knows openrouter has "anthropic/claude-3-5-sonnet")

// Explicit allowedModels with prefix - validates against catalog
mc.IsModelAllowedForProvider("openrouter", "gpt-4o", []string{"openai/gpt-4o"})
// Returns: true (openrouter's catalog contains "openai/gpt-4o" AND model part is "gpt-4o")

// Explicit allowedModels with prefix - wrong model
mc.IsModelAllowedForProvider("openrouter", "claude-3-5-sonnet", []string{"openai/gpt-4o"})
// Returns: false (model part "gpt-4o" doesn't match request "claude-3-5-sonnet")

// Explicit allowedModels without prefix
mc.IsModelAllowedForProvider("openai", "gpt-4o", []string{"gpt-4o"})
// Returns: true (direct match)

func (*ModelCatalog) IsSameModel added in v1.2.19 

func (mc *ModelCatalog) IsSameModel(model1, model2 string) bool

IsSameModel checks if two model strings refer to the same underlying model. It compares the canonical base model names derived from the pricing catalog (or algorithmic fallback for models not in the catalog).

Examples: 

mc.IsSameModel("gpt-4o", "gpt-4o")                            // true (direct match)
mc.IsSameModel("openai/gpt-4o", "gpt-4o")                     // true (same base model)
mc.IsSameModel("gpt-4o", "claude-3-5-sonnet")                  // false (different models)
mc.IsSameModel("openai/gpt-4o", "anthropic/claude-3-5-sonnet") // false

func (*ModelCatalog) IsTextCompletionSupported added in v1.2.11 

func (mc *ModelCatalog) IsTextCompletionSupported(model string, provider schemas.ModelProvider) bool

IsTextCompletionSupported checks if a model supports text completion for the given provider. Returns true if the model has pricing data for text completion ("text_completion"), false otherwise. This is used by the litellmcompat plugin to determine whether to convert text completion requests to chat completion requests.

func (*ModelCatalog) RefineModelForProvider added in v1.1.21 

func (mc *ModelCatalog) RefineModelForProvider(provider schemas.ModelProvider, model string) (string, error)

RefineModelForProvider refines the model for a given provider by performing a lookup in mc.modelPool and using schemas.ParseModelString to extract provider and model parts. e.g. "gpt-oss-120b" for groq provider -> "openai/gpt-oss-120b"

Behavior: - When the provider's catalog (mc.modelPool) yields multiple matching models, returns an error - When exactly one match is found, returns the fully-qualified model (provider/model format) - When the provider is not handled or no refinement is needed, returns the original model unchanged

func (*ModelCatalog) ReloadPricing 

func (mc *ModelCatalog) ReloadPricing(ctx context.Context, config *Config) error

ReloadPricing reloads the model catalog from config

func (*ModelCatalog) SetProviderPricingOverrides added in v1.2.22 

func (mc *ModelCatalog) SetProviderPricingOverrides(provider schemas.ModelProvider, overrides []schemas.ProviderPricingOverride) error

func (*ModelCatalog) UpsertModelDataForProvider added in v1.2.19 

func (mc *ModelCatalog) UpsertModelDataForProvider(provider schemas.ModelProvider, modelData *schemas.BifrostListModelsResponse, allowedModels []schemas.Model, deniedModels []schemas.Model)

UpsertModelDataForProvider upserts model data for a given provider

func (*ModelCatalog) UpsertUnfilteredModelDataForProvider added in v1.2.22 

func (mc *ModelCatalog) UpsertUnfilteredModelDataForProvider(provider schemas.ModelProvider, modelData *schemas.BifrostListModelsResponse)

UpsertUnfilteredModelDataForProvider upserts unfiltered model data for a given provider

type PricingEntry 

type PricingEntry struct {
	BaseModel string `json:"base_model,omitempty"`
	Provider  string `json:"provider"`
	Mode      string `json:"mode"`

	ContextLength   *int                  `json:"context_length,omitempty"`
	MaxInputTokens  *int                  `json:"max_input_tokens,omitempty"`
	MaxOutputTokens *int                  `json:"max_output_tokens,omitempty"`
	Architecture    *schemas.Architecture `json:"architecture,omitempty"`

	// Costs - Text
	InputCostPerToken          float64  `json:"input_cost_per_token"`
	OutputCostPerToken         float64  `json:"output_cost_per_token"`
	InputCostPerTokenBatches   *float64 `json:"input_cost_per_token_batches,omitempty"`
	OutputCostPerTokenBatches  *float64 `json:"output_cost_per_token_batches,omitempty"`
	InputCostPerTokenPriority  *float64 `json:"input_cost_per_token_priority,omitempty"`
	OutputCostPerTokenPriority *float64 `json:"output_cost_per_token_priority,omitempty"`
	InputCostPerCharacter      *float64 `json:"input_cost_per_character,omitempty"`
	// Costs - 128k Tier
	InputCostPerTokenAbove128kTokens          *float64 `json:"input_cost_per_token_above_128k_tokens,omitempty"`
	InputCostPerImageAbove128kTokens          *float64 `json:"input_cost_per_image_above_128k_tokens,omitempty"`
	InputCostPerVideoPerSecondAbove128kTokens *float64 `json:"input_cost_per_video_per_second_above_128k_tokens,omitempty"`
	InputCostPerAudioPerSecondAbove128kTokens *float64 `json:"input_cost_per_audio_per_second_above_128k_tokens,omitempty"`
	OutputCostPerTokenAbove128kTokens         *float64 `json:"output_cost_per_token_above_128k_tokens,omitempty"`
	// Costs - 200k Tier
	InputCostPerTokenAbove200kTokens          *float64 `json:"input_cost_per_token_above_200k_tokens,omitempty"`
	InputCostPerTokenAbove200kTokensPriority  *float64 `json:"input_cost_per_token_above_200k_tokens_priority,omitempty"`
	OutputCostPerTokenAbove200kTokens         *float64 `json:"output_cost_per_token_above_200k_tokens,omitempty"`
	OutputCostPerTokenAbove200kTokensPriority *float64 `json:"output_cost_per_token_above_200k_tokens_priority,omitempty"`
	// Costs - 272k Tier
	InputCostPerTokenAbove272kTokens          *float64 `json:"input_cost_per_token_above_272k_tokens,omitempty"`
	InputCostPerTokenAbove272kTokensPriority  *float64 `json:"input_cost_per_token_above_272k_tokens_priority,omitempty"`
	OutputCostPerTokenAbove272kTokens         *float64 `json:"output_cost_per_token_above_272k_tokens,omitempty"`
	OutputCostPerTokenAbove272kTokensPriority *float64 `json:"output_cost_per_token_above_272k_tokens_priority,omitempty"`

	// Costs - Cache
	CacheCreationInputTokenCost                        *float64 `json:"cache_creation_input_token_cost,omitempty"`
	CacheReadInputTokenCost                            *float64 `json:"cache_read_input_token_cost,omitempty"`
	CacheCreationInputTokenCostAbove200kTokens         *float64 `json:"cache_creation_input_token_cost_above_200k_tokens,omitempty"`
	CacheReadInputTokenCostAbove200kTokens             *float64 `json:"cache_read_input_token_cost_above_200k_tokens,omitempty"`
	CacheReadInputTokenCostAbove200kTokensPriority     *float64 `json:"cache_read_input_token_cost_above_200k_tokens_priority,omitempty"`
	CacheCreationInputTokenCostAbove1hr                *float64 `json:"cache_creation_input_token_cost_above_1hr,omitempty"`
	CacheCreationInputTokenCostAbove1hrAbove200kTokens *float64 `json:"cache_creation_input_token_cost_above_1hr_above_200k_tokens,omitempty"`
	CacheCreationInputAudioTokenCost                   *float64 `json:"cache_creation_input_audio_token_cost,omitempty"`
	CacheReadInputTokenCostPriority                    *float64 `json:"cache_read_input_token_cost_priority,omitempty"`
	CacheReadInputImageTokenCost                       *float64 `json:"cache_read_input_image_token_cost,omitempty"`
	CacheReadInputTokenCostAbove272kTokens             *float64 `json:"cache_read_input_token_cost_above_272k_tokens,omitempty"`
	CacheReadInputTokenCostAbove272kTokensPriority     *float64 `json:"cache_read_input_token_cost_above_272k_tokens_priority,omitempty"`

	// Costs - Image
	InputCostPerImage                             *float64 `json:"input_cost_per_image,omitempty"`
	InputCostPerPixel                             *float64 `json:"input_cost_per_pixel,omitempty"`
	OutputCostPerImage                            *float64 `json:"output_cost_per_image,omitempty"`
	OutputCostPerPixel                            *float64 `json:"output_cost_per_pixel,omitempty"`
	OutputCostPerImagePremiumImage                *float64 `json:"output_cost_per_image_premium_image,omitempty"`
	OutputCostPerImageAbove512x512Pixels          *float64 `json:"output_cost_per_image_above_512_and_512_pixels,omitempty"`
	OutputCostPerImageAbove512x512PixelsPremium   *float64 `json:"output_cost_per_image_above_512_and_512_pixels_and_premium_image,omitempty"`
	OutputCostPerImageAbove1024x1024Pixels        *float64 `json:"output_cost_per_image_above_1024_and_1024_pixels,omitempty"`
	OutputCostPerImageAbove1024x1024PixelsPremium *float64 `json:"output_cost_per_image_above_1024_and_1024_pixels_and_premium_image,omitempty"`
	OutputCostPerImageAbove2048x2048Pixels        *float64 `json:"output_cost_per_image_above_2048_and_2048_pixels,omitempty"`
	OutputCostPerImageAbove4096x4096Pixels        *float64 `json:"output_cost_per_image_above_4096_and_4096_pixels,omitempty"`
	OutputCostPerImageLowQuality                  *float64 `json:"output_cost_per_image_low_quality,omitempty"`
	OutputCostPerImageMediumQuality               *float64 `json:"output_cost_per_image_medium_quality,omitempty"`
	OutputCostPerImageHighQuality                 *float64 `json:"output_cost_per_image_high_quality,omitempty"`
	OutputCostPerImageAutoQuality                 *float64 `json:"output_cost_per_image_auto_quality,omitempty"`
	InputCostPerImageToken                        *float64 `json:"input_cost_per_image_token,omitempty"`
	OutputCostPerImageToken                       *float64 `json:"output_cost_per_image_token,omitempty"`

	// Costs - Audio/Video
	InputCostPerAudioToken      *float64 `json:"input_cost_per_audio_token,omitempty"`
	InputCostPerAudioPerSecond  *float64 `json:"input_cost_per_audio_per_second,omitempty"`
	InputCostPerSecond          *float64 `json:"input_cost_per_second,omitempty"`
	InputCostPerVideoPerSecond  *float64 `json:"input_cost_per_video_per_second,omitempty"`
	OutputCostPerAudioToken     *float64 `json:"output_cost_per_audio_token,omitempty"`
	OutputCostPerVideoPerSecond *float64 `json:"output_cost_per_video_per_second,omitempty"`
	OutputCostPerSecond         *float64 `json:"output_cost_per_second,omitempty"`

	// Costs - Other
	//
	// SearchContextCostPerQuery is stored as a single float64, but the pricing datasheet
	// represents it as a tiered object with three keys: search_context_size_low,
	// search_context_size_medium, and search_context_size_high.  For every provider except
	// Perplexity the three tier values are identical, so we collapse the object to its
	// medium tier value (falling back to low then high).  Perplexity always returns a
	// pre-computed total_cost in its usage response, so the per-query rate is never
	// consumed for that provider; the collapsed value is therefore correct in all cases.
	// See UnmarshalJSON below for the custom decoding logic.
	SearchContextCostPerQuery     *float64 `json:"search_context_cost_per_query,omitempty"`
	CodeInterpreterCostPerSession *float64 `json:"code_interpreter_cost_per_session,omitempty"`
}

PricingEntry represents a single model's pricing information. Field names and JSON tags match the datasheet schema exactly.

func (*PricingEntry) UnmarshalJSON added in v1.2.27 

func (p *PricingEntry) UnmarshalJSON(data []byte) error

UnmarshalJSON implements json.Unmarshaler for PricingEntry. It handles the special case where search_context_cost_per_query may arrive as either a plain float64 or a tiered object {"search_context_size_low":…, "search_context_size_medium":…, "search_context_size_high":…}.

type ShouldSyncPricingFunc added in v1.2.0 

type ShouldSyncPricingFunc func(ctx context.Context) bool

ShouldSyncPricingFunc is a function that determines if pricing data should be synced It returns a boolean indicating if syncing is needed It is completely optional and can be nil if not needed syncPricing function will be called if this function returns true

Source Files

View all Source files

Jump to

Keyboard shortcuts

? : This menu
/ : Search site
f or F : Jump to
y or Y : Canonical URL
go.dev uses cookies from Google to deliver and enhance the quality of its services and to analyze traffic. Learn more.