models

package
v0.9.0 Latest Latest
Warning

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

 Go to latest Published: Oct 29, 2026 License: Apache-2.0, MIT Imports: 4 Imported by: 0

Documentation

Overview

Package models defines shared, credential-free model metadata.

Index

Constants

View Source 
const (
	// ExtraAPI forces a protocol ("openai-responses", "anthropic-messages", ...).
	ExtraAPI = "api"
	// ExtraRemoteModelName sends a different model id on the wire.
	ExtraRemoteModelName = "remote_model_name"
	// ExtraAPIVersion is the Azure api-version query parameter.
	ExtraAPIVersion = "api_version"
	// ExtraThinkingControl is the legacy thinking encoding selector written
	// by older UIs: none | enable_thinking | thinking_type | chat_template_kwargs.
	ExtraThinkingControl = "thinking_control"
	// ExtraTruncatePromptTokens is the vLLM-only server-side truncation
	// budget for rerank, opt-in per row. It is never sent unless the operator
	// set it: vendors that do not implement it reject the unknown field.
	ExtraTruncatePromptTokens = "truncate_prompt_tokens"
	// LegacyTruncatePromptTokens is the embedding truncation budget the
	// pre-catalog OpenAI-compatible client sent whenever a row left it at 0.
	LegacyTruncatePromptTokens = 511
	// ExtraScoreScale overrides the vendor rerank score scale for one row.
	// A self-hosted gateway serves whatever reranker was deployed behind it
	// and normalization depends on the serving configuration, not the
	// model family. Already-normalized 0..1 relevance uses probability;
	// raw logits use logit. A vendor can only state its documented default.
	ExtraScoreScale = "score_scale"
)

Extra-config keys honoured by Resolve. They predate the catalog and stay supported so existing model rows keep working unchanged.

Variables

This section is empty.

Functions

func ParseModelType 

func ParseModelType(s string) (types.ModelType, bool)

ParseModelType accepts both frontend ("chat") and backend ("KnowledgeQA") spellings.

Types

type Capabilities 

type Capabilities struct {
	Provider        string                `json:"provider"`
	API             api.API               `json:"api" extensions:"x-agent"`
	Cataloged       bool                  `json:"cataloged" extensions:"x-agent"`
	Reasoning       bool                  `json:"reasoning" extensions:"x-agent"`
	ThinkingLevels  []api.ReasoningEffort `json:"thinking_levels" extensions:"x-agent"`
	ThinkingFormat  string                `json:"thinking_format"`
	Input           []string              `json:"input,omitempty" extensions:"x-agent"`
	ContextWindow   int                   `json:"context_window,omitempty" extensions:"x-agent"`
	MaxOutputTokens int                   `json:"max_output_tokens,omitempty" extensions:"x-agent"`
	MaxTokensField  string                `json:"max_tokens_field,omitempty"`
}

Capabilities is the UI-facing summary of a resolved model.

type ModelCost 

type ModelCost struct {
	Input      float64 `json:"input"`
	Output     float64 `json:"output"`
	CacheRead  float64 `json:"cache_read,omitempty"`
	CacheWrite float64 `json:"cache_write,omitempty"`
	Currency   string  `json:"currency,omitempty"`
}

ModelCost is priced per million tokens, in USD unless Currency says otherwise.

type ModelSpec 

type ModelSpec struct {
	ID   string `json:"id"`
	Name string `json:"name,omitempty"`
	// Type is the WeKnora model type this entry describes. Chat models with
	// image input are still KnowledgeQA; the UI derives VLM eligibility from
	// Input. Embedding / Rerank / ASR entries exist for the model picker.
	Type types.ModelType `json:"type,omitempty"`
	API  api.API         `json:"api,omitempty"`
	// Aliases are additional ids that resolve to this entry (dated
	// snapshots, vendor-prefixed names on gateways).
	Aliases []string `json:"aliases,omitempty"`
	// Match is a case-insensitive glob ("gpt-5*", "*deepseek-r1*") that lets
	// one entry describe a family. Exact ids and aliases win over patterns;
	// among patterns the longest literal prefix wins.
	Match string `json:"match,omitempty"`
	// Reasoning reports whether the model can emit thinking at all.
	Reasoning bool `json:"reasoning"`
	// Input lists accepted modalities: text, image, audio, video.
	Input           []string   `json:"input,omitempty"`
	ContextWindow   int        `json:"context_window,omitempty"`
	MaxOutputTokens int        `json:"max_output_tokens,omitempty"`
	Cost            *ModelCost `json:"cost,omitempty"`
	// ThinkingLevels maps protocol-neutral levels to vendor values; missing
	// keys inherit the vendor map. See api.ThinkingLevelMap.
	ThinkingLevels api.ThinkingLevelMap `json:"thinking_levels,omitempty"`
	// Compat is the protocol-specific overlay for this model (flat object,
	// fields depend on API). Decoded lazily by Resolve.
	Compat json.RawMessage `json:"compat,omitempty"`
	// Dimension is the embedding vector width (embedding entries only).
	Dimension int `json:"dimension,omitempty"`
	// Deprecated hides the entry from pickers but keeps resolution working.
	Deprecated bool `json:"deprecated,omitempty"`
	// Source is the documentation URL the facts were taken from.
	Source string `json:"source,omitempty"`
}

ModelSpec is one entry of the generated model catalog.

Only ID is required. API defaults to the vendor's API; Type defaults to KnowledgeQA. Compat is interpreted according to API (see compat.go).

func (ModelSpec) AcceptsImages 

func (m ModelSpec) AcceptsImages() bool

AcceptsImages reports whether Input lists image.

func (ModelSpec) DisplayName 

func (m ModelSpec) DisplayName() string

DisplayName returns Name or the id.

Source Files

View all Source files

Directories

Path Synopsis
api
Package api holds the protocol-neutral request model shared by every wire protocol implementation (OpenAI Chat Completions, OpenAI Responses, Anthropic Messages, Google Generative AI, Ollama) plus the helpers those implementations share: SSE reading, the SSRF-safe transport, streaming tool-call assembly, prompt-cache bookkeeping and usage logging.
 Package api holds the protocol-neutral request model shared by every wire protocol implementation (OpenAI Chat Completions, OpenAI Responses, Anthropic Messages, Google Generative AI, Ollama) plus the helpers those implementations share: SSE reading, the SSRF-safe transport, streaming tool-call assembly, prompt-cache bookkeeping and usage logging.
anthropicmessages
Package anthropicmessages implements the Anthropic Messages wire protocol (https://docs.anthropic.com/en/api/messages).
 Package anthropicmessages implements the Anthropic Messages wire protocol (https://docs.anthropic.com/en/api/messages).
arkembeddings
Package arkembeddings implements Volcengine Ark's multimodal embedding shape: an input array of typed parts answering a single `data` object.
 Package arkembeddings implements Volcengine Ark's multimodal embedding shape: an input array of typed parts answering a single `data` object.
cohererank
Package cohererank implements the rerank shape Cohere defined and almost every other vendor copied: POST {base}/rerank with {model, query, documents} answering {results: [{index, relevance_score, document}]}.
 Package cohererank implements the rerank shape Cohere defined and almost every other vendor copied: POST {base}/rerank with {model, query, documents} answering {results: [{index, relevance_score, document}]}.
dashscopeembeddings
Package dashscopeembeddings implements Alibaba Model Studio's native multimodal embedding shape: input.contents, parameters.dimension, and results under output.embeddings.
 Package dashscopeembeddings implements Alibaba Model Studio's native multimodal embedding shape: input.contents, parameters.dimension, and results under output.embeddings.
dashscoperank
Package dashscoperank implements Alibaba Model Studio's native text-rerank shape: the same fields as the Cohere dialect, wrapped in input/parameters and answered under output.
 Package dashscoperank implements Alibaba Model Studio's native text-rerank shape: the same fields as the Cohere dialect, wrapped in input/parameters and answered under output.
googleembeddings
Package googleembeddings implements Gemini's batchEmbedContents: one POST carrying a request per text, answering embeddings[].values in order.
 Package googleembeddings implements Gemini's batchEmbedContents: one POST carrying a request per text, answering embeddings[].values in order.
googlegenai
Package googlegenai implements the native Google Gemini generateContent wire protocol (https://ai.google.dev/api/generate-content).
 Package googlegenai implements the native Google Gemini generateContent wire protocol (https://ai.google.dev/api/generate-content).
nimrerank
Package nimrerank implements NVIDIA NIM's retrieval reranking shape: a query object, a passages array, and rankings carrying a logit.
 Package nimrerank implements NVIDIA NIM's retrieval reranking shape: a query object, a passages array, and rankings carrying a logit.
openaichataudio
Package openaichataudio implements speech recognition served through the chat completions endpoint: one user message whose only content part is {"type": "input_audio", "input_audio": {"data": "data:<mime>;base64,..."}}, answered by an assistant message whose content is the transcript.
 Package openaichataudio implements speech recognition served through the chat completions endpoint: one user message whose only content part is {"type": "input_audio", "input_audio": {"data": "data:<mime>;base64,..."}}, answered by an assistant message whose content is the transcript.
openaicompletions
Package openaicompletions implements the OpenAI Chat Completions wire protocol.
 Package openaicompletions implements the OpenAI Chat Completions wire protocol.
openaiembeddings
Package openaiembeddings implements the embedding shape OpenAI defined and every compatible gateway copied: POST {base}/embeddings with {model, input} answering {data: [{embedding, index}]}.
 Package openaiembeddings implements the embedding shape OpenAI defined and every compatible gateway copied: POST {base}/embeddings with {model, input} answering {data: [{embedding, index}]}.
openairesponses
Package openairesponses implements the OpenAI Responses wire protocol (POST /responses).
 Package openairesponses implements the OpenAI Responses wire protocol (POST /responses).
openaitranscriptions
Package openaitranscriptions implements the speech-to-text shape OpenAI defined and the compatible servers copied: a multipart POST to {base}/audio/transcriptions carrying file and model, answering {text}, or {text, segments} when response_format is verbose_json.
 Package openaitranscriptions implements the speech-to-text shape OpenAI defined and the compatible servers copied: a multipart POST to {base}/audio/transcriptions carrying file and model, answering {text}, or {text, segments} when response_format is verbose_json.
pineconerank
Package pineconerank implements Pinecone Inference's standalone rerank API.
 Package pineconerank implements Pinecone Inference's standalone rerank API.
teirank
Package teirank implements the native Text Embeddings Inference rerank protocol.
 Package teirank implements the native Text Embeddings Inference rerank protocol.
tencentlkeap
Package tencentlkeap implements Tencent RunRerank through its official SDK.
 Package tencentlkeap implements Tencent RunRerank through its official SDK.
volcengineknowledge
Package volcengineknowledge implements Volcengine Knowledge Service rerank through its official SDK.
 Package volcengineknowledge implements Volcengine Knowledge Service rerank through its official SDK.
Package catalog loads and queries model metadata.
 Package catalog loads and queries model metadata.
Package chat exposes the Chat interface used by the application layer and builds the right protocol client for a configured model.
 Package chat exposes the Chat interface used by the application layer and builds the right protocol client for a configured model.
Package imageprep turns stored image bytes into an image a model will take: converted to a format the vendor documents and shrunk to its size limit.
 Package imageprep turns stored image bytes into an image a model will take: converted to a format the vendor documents and shrunk to its size limit.
internal
configcopy
Package configcopy copies exported configuration data and immutable function values.
 Package configcopy copies exported configuration data and immutable function values.
Package limiter provides a distributed, per-key concurrency governor for outbound model-provider calls.
 Package limiter provides a distributed, per-key concurrency governor for outbound model-provider calls.
Package providers registers Alibaba Cloud Model Studio (Bailian / DashScope) through its OpenAI-compatible mode.
 Package providers registers Alibaba Cloud Model Studio (Bailian / DashScope) through its OpenAI-compatible mode.
Package rerank adapts rerank protocols to application batching and score semantics.
 Package rerank adapts rerank protocols to application batching and score semantics.
Package runtime resolves one model's effective settings and assembles its authenticated endpoint.
 Package runtime resolves one model's effective settings and assembles its authenticated endpoint.
ollama

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.