openrouter

type ChatConfig struct {
	compat_oai.RequestConfig

	// Temperature controls the degree of randomness in token selection, from
	// 0 to 2.
	Temperature *float64 `` /* 153-byte string literal not displayed */
	// TopP is the nucleus sampling threshold, from 0 to 1.
	TopP *float64 `` /* 190-byte string literal not displayed */
	// TopK limits sampling to the K most likely tokens, sent as the API's
	// top_k. 0, the default, applies no limit. Some vendors ignore it.
	TopK *int `` /* 173-byte string literal not displayed */
	// MaxOutputTokens is the maximum number of tokens to generate, sent as
	// the API's max_tokens.
	MaxOutputTokens int `` /* 148-byte string literal not displayed */
	// StopSequences stop generation when produced by the model, up to four.
	StopSequences []string `` /* 145-byte string literal not displayed */
	// FrequencyPenalty penalizes tokens by their frequency so far, from -2.0
	// to 2.0.
	FrequencyPenalty *float64 `` /* 154-byte string literal not displayed */
	// PresencePenalty penalizes tokens that have appeared at all, from -2.0
	// to 2.0.
	PresencePenalty *float64 `` /* 153-byte string literal not displayed */
	// RepetitionPenalty penalizes tokens by whether they appeared in the
	// input, from 0 to 2, sent as the API's repetition_penalty. 1 is neutral.
	RepetitionPenalty *float64 `` /* 213-byte string literal not displayed */
	// MinP is the minimum probability a token needs relative to the most
	// likely one, from 0 to 1, sent as the API's min_p.
	MinP *float64 `` /* 184-byte string literal not displayed */
	// TopA filters tokens by a threshold scaled from the most likely token's
	// probability, from 0 to 1, sent as the API's top_a.
	TopA *float64 `` /* 196-byte string literal not displayed */
	// Seed makes generation reproducible across calls when set, on a
	// best-effort basis.
	Seed *int `json:"seed,omitempty" jsonschema_description:"Makes generation reproducible across calls when set, on a best-effort basis."`
	// LogProbs requests log probabilities for the output tokens.
	LogProbs *bool `json:"logProbs,omitempty" jsonschema_description:"Requests log probabilities for the output tokens."`
	// TopLogProbs is how many of the most likely tokens to return log
	// probabilities for at each position, from 0 to 20; it requires LogProbs.
	TopLogProbs *int `` /* 205-byte string literal not displayed */
	// ParallelToolCalls lets the model request several tool calls in one
	// response, which it may do by default. It applies to a request that
	// carries tools.
	ParallelToolCalls *bool `` /* 186-byte string literal not displayed */
	// User identifies the end user a request is made for, which OpenRouter
	// uses to isolate abuse to one user rather than the whole key.
	User string `` /* 171-byte string literal not displayed */

	// Models lists further models to fall back to, in order, when the
	// requested one is unavailable, rate-limited, or refuses. The model the
	// request names is tried first.
	Models []string `` /* 191-byte string literal not displayed */
	// Provider chooses which upstream providers may serve the request.
	Provider *ProviderRouting `json:"provider,omitempty" jsonschema_description:"Chooses which upstream providers may serve the request, and in what order."`
	// Reasoning controls the reasoning the model does before answering.
	Reasoning *ReasoningConfig `json:"reasoning,omitempty" jsonschema_description:"Controls the reasoning the model does before answering."`
	// Plugins enables OpenRouter's request plugins, such as web search and
	// file parsing, each an object with an id and that plugin's own options:
	//
	//	Plugins: []map[string]any{{"id": "web", "max_results": 3}}
	//
	// They are sent verbatim rather than typed: the roster and each plugin's
	// options change on OpenRouter's schedule, and a struct here would reject
	// an option the gateway accepts. See
	// https://openrouter.ai/docs/features/web-search.
	Plugins []map[string]any `` /* 214-byte string literal not displayed */
	// Transforms are the prompt transforms to apply, currently "middle-out",
	// which compresses a prompt that would overflow the model's context by
	// dropping from the middle.
	Transforms []string `` /* 197-byte string literal not displayed */
	// SessionID groups related requests so they keep reaching the same
	// upstream provider, sent as the API's session_id. It is what keeps a
	// multi-turn conversation on one provider's prompt cache.
	SessionID string `` /* 222-byte string literal not displayed */
	// ServiceTier selects how the request is scheduled and billed upstream.
	ServiceTier ServiceTier `` /* 231-byte string literal not displayed */
	// Metadata attaches key-value pairs to the request, readable later on
	// OpenRouter's activity pages. It takes up to 16 pairs, with keys up to
	// 64 characters and values up to 512.
	Metadata map[string]string `` /* 201-byte string literal not displayed */
}

type DataCollection string

type MaxPrice struct {
	// Prompt is the cap on the input price, per million tokens.
	Prompt *float64 `json:"prompt,omitempty" jsonschema:"minimum=0" jsonschema_description:"Cap on the input price, in USD per million tokens."`
	// Completion is the cap on the output price, per million tokens.
	Completion *float64 `` /* 127-byte string literal not displayed */
	// Request is the cap on the per-request price.
	Request *float64 `json:"request,omitempty" jsonschema:"minimum=0" jsonschema_description:"Cap on the per-request price, in USD."`
	// Image is the cap on the per-image price.
	Image *float64 `json:"image,omitempty" jsonschema:"minimum=0" jsonschema_description:"Cap on the per-image price, in USD."`
}

type OpenRouter struct {
	// APIKey is the OpenRouter API key. If empty, OPENROUTER_API_KEY is
	// consulted.
	APIKey string
	// SiteURL identifies the calling application on OpenRouter's rankings,
	// sent as the HTTP-Referer header. It is optional and affects
	// attribution only.
	SiteURL string
	// AppName is the title the calling application appears under on
	// OpenRouter's rankings, sent as the X-Title header. It is optional and
	// affects attribution only.
	AppName string
	// Opts contains additional OpenAI client request options, such as
	// [option.WithBaseURL] for a different endpoint (OPENROUTER_BASE_URL
	// works too). Options supplied here are applied after the plugin
	// defaults, so they win on overlap.
	Opts []option.RequestOption

	// Models overrides what the plugin knows about a model, keyed by model
	// ID, bare or provider-prefixed. Every model already works without an
	// entry: OpenRouter serves hundreds of models from dozens of vendors and
	// adds more weekly, so the plugin curates no catalog and describes every
	// model it resolves with the same deliberately permissive capabilities.
	// The two ways to be wrong are not symmetric: a capability declared too
	// narrow is refused by Genkit before the request is sent, which blocks a
	// model that would have worked, while one declared too wide reaches
	// OpenRouter, which answers with the real reason the model cannot serve
	// it. Constrained output is the exception, left unclaimed on purpose:
	// a large share of the catalog lacks it natively, and unset, Genkit
	// falls back to putting the schema in the prompt, which every model
	// handles and which returns the same typed result.
	//
	// Supply an entry to correct what a model can actually do:
	//
	//	&openrouter.OpenRouter{Models: map[string]ai.ModelOptions{
	//		// A text-only model, so Genkit refuses media locally rather
	//		// than paying for the upstream rejection.
	//		"mistralai/mistral-7b-instruct": {Supports: &ai.ModelSupports{
	//			Multiturn: true, Tools: true, SystemRole: true,
	//		}},
	//	}}
	//
	// Fields left at their zero value keep what the plugin resolves, so an
	// entry can pin one capability without restating the rest. The model ID
	// keeps the upstream vendor's prefix, so it contains a slash; the
	// optional provider prefix is this plugin's own, as in
	// "openrouter/mistralai/mistral-7b-instruct".
	Models map[string]ai.ModelOptions
	// contains filtered or unexported fields
}

type ProviderRouting struct {
	// Order lists provider slugs to try in order before any fallback.
	Order []string `` /* 127-byte string literal not displayed */
	// Only restricts routing to these provider slugs.
	Only []string `json:"only,omitempty" jsonschema_description:"Restricts routing to these provider slugs."`
	// Ignore skips these provider slugs.
	Ignore []string `json:"ignore,omitempty" jsonschema_description:"Skips these provider slugs."`
	// AllowFallbacks lets another provider serve the request when the chosen
	// ones fail. It defaults to true; false fails the request instead.
	AllowFallbacks *bool `` /* 174-byte string literal not displayed */
	// RequireParameters routes only to providers that honor every parameter
	// the request carries, rather than to one that would ignore some.
	RequireParameters *bool `` /* 175-byte string literal not displayed */
	// DataCollection restricts routing to providers that do not store request
	// data when set to deny.
	DataCollection DataCollection `` /* 189-byte string literal not displayed */
	// ZDR restricts routing to zero data retention endpoints.
	ZDR *bool `json:"zdr,omitempty" jsonschema_description:"Restricts routing to zero data retention endpoints."`
	// Sort ranks providers by one metric instead of OpenRouter's default load
	// balancing.
	Sort ProviderSort `` /* 187-byte string literal not displayed */
	// Quantizations restricts routing to providers serving the model at these
	// quantization levels, e.g. int4, int8, fp8, fp16, bf16, or fp32. The
	// set is not enumerated in the schema: OpenRouter adds levels as hardware
	// gains them, and a closed list here would reject a level it accepts.
	Quantizations []string `` /* 177-byte string literal not displayed */
	// MaxPrice caps what the request may cost.
	MaxPrice *MaxPrice `` /* 177-byte string literal not displayed */
	// PreferredMinThroughput deprioritizes providers below this many output
	// tokens per second. They stay eligible as a fallback.
	PreferredMinThroughput *float64 `` /* 187-byte string literal not displayed */
	// PreferredMaxLatency deprioritizes providers slower than this many
	// seconds to first token. They stay eligible as a fallback.
	PreferredMaxLatency *float64 `` /* 188-byte string literal not displayed */
}

type ProviderSort string

type ReasoningConfig struct {
	// Effort is how much reasoning to spend, from none to max. Vendors that
	// budget in tokens rather than levels take MaxTokens instead.
	Effort ReasoningEffort `` /* 220-byte string literal not displayed */
	// MaxTokens is the exact token budget to reason with, sent as the API's
	// max_tokens inside the reasoning object. It overrides Effort. Vendors
	// that take a budget usually reject one under 1,024 tokens, which is a
	// per-vendor limit rather than a documented API-wide one.
	MaxTokens int `` /* 199-byte string literal not displayed */
	// Exclude keeps the reasoning out of the response without stopping the
	// model from reasoning.
	Exclude *bool `` /* 132-byte string literal not displayed */
	// Enabled turns reasoning on with the vendor's own defaults, which
	// OpenRouter treats as medium effort. Effort and MaxTokens imply it.
	Enabled *bool `` /* 169-byte string literal not displayed */
}

type ReasoningEffort string

type ServiceTier string