model

package
v0.14.5 Latest Latest
Warning

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

 Go to latest Published: Aug 15, 2026 License: MIT Imports: 2 Imported by: 0

Documentation

Overview

Package model is the single source of truth for the models wingman serves: display name, capability class and token limits. Values are kept in sync with the wingman-vscode catalog (src/models.ts).

Index

Constants

This section is empty.

Variables

View Source 
var Models = []Model{

	{
		ID: "claude-sonnet-5",

		Name:  "Claude Sonnet 5",
		Class: ClassMedium,

		Context: 1000000,
		Output:  128000,
	},
	{
		ID: "claude-opus-5",

		Name:  "Claude Opus 5",
		Class: ClassLarge,

		Context: 1000000,
		Output:  128000,
	},

	{
		ID: "claude-opus-4-8",

		Name:  "Claude Opus 4.8",
		Class: ClassLarge,

		Context: 1000000,
		Output:  128000,
	},

	{
		ID: "claude-opus-4-7",

		Name:  "Claude Opus 4.7",
		Class: ClassLarge,

		Context: 1000000,
		Output:  128000,
	},

	{
		ID: "claude-sonnet-4-6",

		Name:  "Claude Sonnet 4.6",
		Class: ClassMedium,

		Context: 1000000,
		Output:  128000,
	},
	{
		ID: "claude-opus-4-6",

		Name:  "Claude Opus 4.6",
		Class: ClassLarge,

		Context: 1000000,
		Output:  128000,
	},
	{
		ID: "claude-haiku-4-6",

		Name:  "Claude Haiku 4.6",
		Class: ClassSmall,

		Context: 200000,
		Output:  64000,
	},

	{
		ID: "claude-sonnet-4-5",

		Name:  "Claude Sonnet 4.5",
		Class: ClassMedium,

		Context: 200000,
		Output:  64000,
	},
	{
		ID: "claude-opus-4-5",

		Name:  "Claude Opus 4.5",
		Class: ClassLarge,

		Context: 200000,
		Output:  64000,
	},
	{
		ID: "claude-haiku-4-5",

		Name:  "Claude Haiku 4.5",
		Class: ClassSmall,

		Context: 200000,
		Output:  64000,
	},

	{
		ID: "claude-fable-5",

		Name:  "Claude Fable 5",
		Class: ClassLarge,

		Context: 1000000,
		Output:  128000,
	},
	{
		ID: "claude-mythos-5",

		Name:  "Claude Mythos 5",
		Class: ClassLarge,

		Context: 1000000,
		Output:  128000,
	},

	{
		ID:      "gpt-5.6-sol",
		Aliases: []string{"gpt-5.6"},

		Name:  "GPT 5.6 Sol",
		Class: ClassLarge,

		Context: 1050000,
		Output:  128000,
	},
	{
		ID: "gpt-5.6-terra",

		Name:  "GPT 5.6 Terra",
		Class: ClassMedium,

		Context: 1050000,
		Output:  128000,
	},
	{
		ID: "gpt-5.6-luna",

		Name:  "GPT 5.6 Luna",
		Class: ClassSmall,

		Context: 1050000,
		Output:  128000,
	},

	{
		ID: "gpt-5.5",

		Name:  "GPT 5.5",
		Class: ClassMedium,

		Context: 1050000,
		Output:  128000,
	},

	{
		ID: "gpt-5.4",

		Name:  "GPT 5.4",
		Class: ClassMedium,

		Context: 1050000,
		Output:  128000,
	},
	{
		ID: "gpt-5.4-mini",

		Name:  "GPT 5.4 Mini",
		Class: ClassSmall,

		Context: 400000,
		Output:  128000,
	},

	{
		ID: "gpt-5.3-codex",

		Name:  "GPT 5.3 Codex",
		Class: ClassMedium,

		Context: 400000,
		Output:  128000,
	},
	{
		ID: "gpt-5.3-codex-spark",

		Name:  "GPT 5.3 Codex Spark",
		Class: ClassSmall,

		Context: 128000,
		Output:  32000,
	},

	{
		ID: "gpt-5.2-codex",

		Name:  "GPT 5.2 Codex",
		Class: ClassMedium,

		Context: 400000,
		Output:  128000,
	},
	{
		ID: "gpt-5.2",

		Name:  "GPT 5.2",
		Class: ClassMedium,

		Context: 400000,
		Output:  128000,
	},

	{
		ID: "gpt-5.1-codex-max",

		Name:  "GPT 5.1 Codex Max",
		Class: ClassMedium,

		Context: 400000,
		Output:  128000,
	},
	{
		ID: "gpt-5.1-codex",

		Name:  "GPT 5.1 Codex",
		Class: ClassMedium,

		Context: 400000,
		Output:  128000,
	},
	{
		ID: "gpt-5.1-codex-mini",

		Name:  "GPT 5.1 Codex Mini",
		Class: ClassSmall,

		Context: 400000,
		Output:  128000,
	},
	{
		ID: "gpt-5.1",

		Name:  "GPT 5.1",
		Class: ClassMedium,

		Context: 400000,
		Output:  128000,
	},

	{
		ID: "gpt-5-codex",

		Name:  "GPT 5 Codex",
		Class: ClassMedium,

		Context: 400000,
		Output:  128000,
	},
	{
		ID: "gpt-5",

		Name:  "GPT 5",
		Class: ClassMedium,

		Context: 400000,
		Output:  128000,
	},
	{
		ID: "gpt-5-mini",

		Name:  "GPT 5 Mini",
		Class: ClassSmall,

		Context: 400000,
		Output:  128000,
	},

	{
		ID: "gemini-3.6-flash",

		Name:  "Gemini 3.6 Flash",
		Class: ClassMedium,

		Input:  1048576,
		Output: 65536,
	},

	{
		ID:      "gemini-3.5-flash",
		Aliases: []string{"gemini-3-flash-preview"},

		Name:  "Gemini 3.5 Flash",
		Class: ClassMedium,

		Input:  1048576,
		Output: 65536,
	},
	{
		ID: "gemini-3.5-flash-lite",

		Name:  "Gemini 3.5 Flash Lite",
		Class: ClassSmall,

		Input:  1048576,
		Output: 65536,
	},

	{
		ID:      "gemini-3.1-pro",
		Aliases: []string{"gemini-3.1-pro-preview"},

		Name:  "Gemini 3.1 Pro",
		Class: ClassLarge,

		Input:  1048576,
		Output: 65536,
	},
	{
		ID: "gemini-3.1-flash-lite",

		Name:  "Gemini 3.1 Flash Lite",
		Class: ClassSmall,

		Input:  1048576,
		Output: 65536,
	},

	{
		ID: "glm-5.3",

		Name:  "GLM 5.3",
		Class: ClassMedium,

		Context: 1000000,
		Output:  131072,
	},

	{
		ID: "glm-5.2",

		Name:  "GLM 5.2",
		Class: ClassMedium,

		Context: 1000000,
		Output:  131072,
	},

	{
		ID: "glm-5.1",

		Name:  "GLM 5.1",
		Class: ClassMedium,

		Context: 200000,
		Output:  131072,
	},

	{
		ID: "glm-5",

		Name:  "GLM 5",
		Class: ClassMedium,

		Context: 204800,
		Output:  131072,
	},

	{
		ID: "glm-4.7",

		Name:  "GLM 4.7",
		Class: ClassMedium,

		Context: 204800,
		Output:  131072,
	},
	{
		ID: "glm-4.7-flash",

		Name:  "GLM 4.7 Flash",
		Class: ClassSmall,

		Context: 200000,
		Output:  131072,
	},

	{
		ID: "deepseek-v4-pro",

		Name:  "DeepSeek V4 Pro",
		Class: ClassLarge,

		Context: 1000000,
		Output:  384000,
	},
	{
		ID: "deepseek-v4-flash",

		Name:  "DeepSeek V4 Flash",
		Class: ClassSmall,

		Context: 1000000,
		Output:  384000,
	},

	{
		ID:      "mistral-medium",
		Aliases: []string{"mistral-medium-latest", "mistral-medium-2604"},

		Name:  "Mistral Medium 3.5",
		Class: ClassMedium,

		Context: 262144,
		Output:  262144,
	},
	{
		ID:      "mistral-small",
		Aliases: []string{"mistral-small-latest", "mistral-small-2603"},

		Name:  "Mistral Small 4",
		Class: ClassSmall,

		Context: 256000,
		Output:  256000,
	},

	{
		ID: "kimi-k3",

		Name:        "Kimi K3",
		Description: "Kimi’s most capable model to date, with 2.8 trillion parameters, native visual understanding, and a 1M-token context window, designed for frontier intelligence scenarios such as software engineering, knowledge work, and deep reasoning.",
		Class:       ClassLarge,

		Context: 1048576,
		Output:  131072,
	},

	{
		ID: "kimi-k2.7-code",

		Name:        "Kimi K2.7 Code",
		Description: "Kimi’s dedicated coding model. It follows instructions more reliably in long contexts, completes coding tasks with higher success rates. Context 256k.",
		Class:       ClassMedium,

		Context: 262144,
		Output:  262144,
	},

	{
		ID:      "minimax-m3",
		Aliases: []string{"MiniMax-M3"},

		Name:  "MiniMax M3",
		Class: ClassLarge,

		Context: 1000000,
		Output:  128000,
	},

	{
		ID: "grok-4.6",

		Name:  "Grok 4.6",
		Class: ClassLarge,

		Context: 500000,
		Output:  500000,
	},

	{
		ID: "qwen3.7-max",

		Name:  "Qwen 3.7 Max",
		Class: ClassLarge,

		Context: 1000000,
		Output:  65536,
	},
	{
		ID:      "qwen3.7-plus",
		Aliases: []string{"qwen3.7"},

		Name:  "Qwen 3.7 Plus",
		Class: ClassMedium,

		Context: 1000000,
		Output:  65536,
	},

	{
		ID:      "qwen3.6-plus",
		Aliases: []string{"qwen3.6"},

		Name:  "Qwen 3.6",
		Class: ClassMedium,

		Context: 1000000,
		Output:  65536,
	},
	{
		ID: "qwen3.6-flash",

		Name:  "Qwen 3.6 Flash",
		Class: ClassSmall,

		Context: 1000000,
		Output:  65536,
	},

	{
		ID:      "qwen3.5-plus",
		Aliases: []string{"qwen3.5"},

		Name:  "Qwen 3.5",
		Class: ClassMedium,

		Context: 1000000,
		Output:  65536,
	},

	{
		ID:      "qwen3-coder-plus",
		Aliases: []string{"qwen3-coder-next", "qwen3-coder"},

		Name:  "Qwen 3 Coder",
		Class: ClassMedium,

		Context: 1048576,
		Output:  65536,
	},
	{
		ID: "qwen3-coder-flash",

		Name:  "Qwen 3 Coder Flash",
		Class: ClassSmall,

		Context: 1048576,
		Output:  65536,
	},
	{
		ID:      "qwen3-next",
		Aliases: []string{"qwen3"},

		Name:  "Qwen 3",
		Class: ClassMedium,

		Input:   126976,
		Output:  32768,
		Context: 131072,
	},
}

Models lists every known model in preference order: the first entry of a family, filter or class is the one picked automatically.

Functions

func Family 

func Family(id string) string

Family groups models by vendor line (claude, gpt, glm, …) so automatic selection stays within one family when possible: switching families mid-session drops encrypted reasoning state.

func Name 

func Name(id string) string

func Normalize added in v0.14.3 

func Normalize(id string) string

Types

type Class 

type Class int

Class buckets models by capability for automatic per-role selection: large drives planning, medium drives coding, small drives utility calls (recaps, compaction summaries) and the "fast" model of external agents. 

const (
	ClassMedium Class = iota
	ClassLarge
	ClassSmall
)

func ClassOf 

func ClassOf(id string) Class

ClassOf reports the class of a model, falling back to naming markers for ids the catalog does not know (external agents report their own models).

type Model 

type Model struct {
	Name        string
	Description string
	Class       Class

	ID string

	Aliases []string

	Input  int
	Output int

	Context int
}

func Available 

func Available(available map[string]bool) []Model

Available returns the catalog entries a backend serves, in preference order and with each ID resolved to the alias the backend accepts. A nil map means availability is unknown and every model is returned.

func Find 

func Find(id string) (Model, bool)

func (Model) ContextTokens 

func (m Model) ContextTokens() int

func (Model) InputTokens 

func (m Model) InputTokens() int

func (Model) OutputTokens 

func (m Model) OutputTokens() int

type Profile added in v0.14.3 

type Profile struct {
	Window                   int
	WindowCompaction         int
	Efforts                  []string
	DefaultEffort            string
	ReasoningEffortPlacement ReasoningEffortPlacement
}

Profile captures per-model harness constraints looked up by id prefix. Window is the full hardware context window. WindowCompaction is the smaller budget compaction runs against by default where long context carries a price premium (0 = compact against Window). Efforts lists the supported reasoning efforts sorted low→high; empty means unrestricted. DefaultEffort overrides the harness role default. ReasoningEffortPlacement describes the provider's JSON request shape.

Verified 2026-07: current Claude models (Opus 4.6+, Sonnet 4.6+, Fable 5) take 1M input tokens at flat per-token rates — no long-context premium; Haiku and pre-4.6 models are 200k hardware (Sonnet 4.5's 1M beta bills 2x above 200k and needs a beta header, so it stays capped). GPT-5.4/5.5 have 1M-class windows but bill 2x input / 1.5x output for the whole session once input exceeds 272k; GPT-5.6 (sol/terra/luna) keeps the same threshold. Codex and earlier GPT-5.x are 400k total, flat. Gemini and Grok bill more above 200k prompts. MiniMax M3's long-context pricing starts above 512k.

func ProfileFor added in v0.14.3 

func ProfileFor(id string) Profile

func (Profile) ContextWindow added in v0.14.3 

func (p Profile) ContextWindow(large bool) int

type ReasoningEffortPlacement added in v0.14.3 

type ReasoningEffortPlacement int
const (
	ReasoningEffortInObject ReasoningEffortPlacement = iota
	ReasoningEffortAtRoot
)

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.