models

const (
	// AnthropicModelsURL is the official Anthropic models documentation page.
	// Use this to check for new models, deprecations, and model updates.
	AnthropicModelsURL = "https://platform.claude.com/docs/en/about-claude/models/overview"

	// AnthropicAPIURL is the Anthropic API reference page.
	AnthropicAPIURL = "https://docs.anthropic.com/en/api"
)

const (
	ClaudeOpus4_5   = "claude-opus-4-5-20251101"   // Claude Opus 4.5 (November 2025)
	ClaudeSonnet4_5 = "claude-sonnet-4-5-20250929" // Claude Sonnet 4.5 (September 2025)
	ClaudeHaiku4_5  = "claude-haiku-4-5-20251001"  // Claude Haiku 4.5 (October 2025)
)

const (
	ClaudeOpus4_1 = "claude-opus-4-1-20250805" // Claude Opus 4.1 (August 2025)
	ClaudeOpus4   = "claude-opus-4-20250514"   // Claude Opus 4 (May 2025)
)

const (
	Claude3Opus   = "claude-3-opus-20240229"   // Claude 3 Opus (February 2024)
	Claude3Sonnet = "claude-3-sonnet-20240229" // Claude 3 Sonnet (February 2024)
	Claude3Haiku  = "claude-3-haiku-20240307"  // Claude 3 Haiku (March 2024)
)

const (
	// BedrockModelsURL is the official AWS Bedrock models documentation page.
	// Use this to check for new models, deprecations, and model updates.
	BedrockModelsURL = "https://docs.aws.amazon.com/bedrock/latest/userguide/models-supported.html"

	// BedrockAPIURL is the AWS Bedrock API reference page.
	BedrockAPIURL = "https://docs.aws.amazon.com/bedrock/latest/APIReference/welcome.html"
)

const (
	// BedrockClaudeOpus4 is Claude Opus 4 on AWS Bedrock.
	BedrockClaudeOpus4 = "anthropic.claude-opus-4-20250514-v1:0"

	// BedrockClaude3Opus is Claude 3 Opus on AWS Bedrock.
	BedrockClaude3Opus = "anthropic.claude-3-opus-20240229-v1:0"

	// BedrockClaude3Sonnet is Claude 3 Sonnet on AWS Bedrock.
	BedrockClaude3Sonnet = "anthropic.claude-3-sonnet-20240229-v1:0"
)

const (
	// GeminiModelsURL is the official Google Gemini models documentation page.
	// Use this to check for new models, deprecations, and model updates.
	GeminiModelsURL = "https://ai.google.dev/gemini-api/docs/models/gemini"

	// GeminiAPIURL is the Google Gemini API reference page.
	GeminiAPIURL = "https://ai.google.dev/gemini-api/docs"
)

const (
	// Gemini2_5Pro is stable with advanced reasoning capabilities.
	Gemini2_5Pro = "gemini-2.5-pro"

	// Gemini2_5Flash is stable with balanced performance.
	Gemini2_5Flash = "gemini-2.5-flash"

	// GeminiLive2_5Flash is the stable Live API model (private GA).
	GeminiLive2_5Flash = "gemini-live-2.5-flash"
)

const (
	Gemini1_5Pro   = "gemini-1.5-pro"   // Gemini 1.5 Pro
	Gemini1_5Flash = "gemini-1.5-flash" // Gemini 1.5 Flash
)

const (
	// GLMModelsURL is the official Zhipu AI models documentation page.
	// Use this to check for new models, deprecations, and model updates.
	GLMModelsURL = "https://bigmodel.cn/dev/api/normal-model/glm-4"

	// GLMAPIURL is the GLM API reference page.
	GLMAPIURL = "https://open.bigmodel.cn/dev/howuse/introduction"
)

const (
	// GLM5 is the flagship model with MoE architecture (744B/40B active).
	// Agentic engineering, 200K context, forced thinking mode.
	GLM5 = "glm-5"

	// GLM5Code is the code-specialized variant of GLM-5.
	// Optimized for programming, 200K context, forced thinking.
	GLM5Code = "glm-5-code"
)

const (
	// GLM4_7 is the premium model with Interleaved Thinking, 200K context.
	GLM4_7 = "glm-4.7"

	// GLM4_7FlashX is the high-speed paid version with priority GPU access.
	// 200K context, hybrid thinking, best price/performance for batch tasks.
	GLM4_7FlashX = "glm-4.7-flashx"

	// GLM4_7Flash is a free SOTA model with 200K context and hybrid thinking.
	// 1 concurrent request limit, ideal for prototyping.
	GLM4_7Flash = "glm-4.7-flash"
)

const (
	// GLM4_5 is the first unified model with reasoning/coding/agent capabilities.
	// MoE 355B/32B active, 128K context, auto-thinking.
	GLM4_5 = "glm-4.5"

	// GLM4_5X is the ultra-fast premium version with lowest latency.
	// 128K context, auto-thinking.
	GLM4_5X = "glm-4.5-x"

	// GLM4_5Air is the cost-effective lightweight model.
	// MoE 106B/12B active, 128K context, auto-thinking.
	GLM4_5Air = "glm-4.5-air"

	// GLM4_5AirX is the accelerated Air version with priority GPU access.
	GLM4_5AirX = "glm-4.5-airx"

	// GLM4_5Flash is a free model with reasoning/coding/agents support.
	// 128K context, auto-thinking, function calling enabled.
	GLM4_5Flash = "glm-4.5-flash"
)

const (
	// KimiModelsURL is the official Moonshot AI models documentation page.
	// Use this to check for new models, deprecations, and model updates.
	KimiModelsURL = "https://platform.moonshot.cn/docs/api/chat"

	// KimiAPIURL is the Kimi API reference page.
	KimiAPIURL = "https://platform.moonshot.cn/docs"
)

const (
	// KimiK2_0905 is an enhanced agentic coding model with improved frontend code
	// quality and context understanding, 256k context window.
	KimiK2_0905 = "kimi-k2-0905-preview"

	// KimiK2_0711 is the MoE base model with powerful code and agent capabilities,
	// 128k context window.
	KimiK2_0711 = "kimi-k2-0711-preview"

	// KimiK2Turbo is the high-speed version of K2-0905, 60-100 tokens/sec output speed,
	// 256k context window.
	KimiK2Turbo = "kimi-k2-turbo-preview"

	// KimiK2Thinking is a long-term thinking model with multi-step tool usage
	// and deep reasoning, 256k context window.
	KimiK2Thinking = "kimi-k2-thinking"

	// KimiK2ThinkingTurbo is the high-speed thinking model, 60-100 tokens/sec,
	// excels at deep reasoning, 256k context window.
	KimiK2ThinkingTurbo = "kimi-k2-thinking-turbo"
)

const (
	// MoonshotV1_8K is suitable for generating short texts with 8k context window.
	MoonshotV1_8K = "moonshot-v1-8k"

	// MoonshotV1_32K is suitable for generating long texts with 32k context window.
	MoonshotV1_32K = "moonshot-v1-32k"

	// MoonshotV1_128K is suitable for generating very long texts with 128k context window.
	MoonshotV1_128K = "moonshot-v1-128k"
)

const (
	// MoonshotV1_8KVision understands image content and outputs text, 8k context.
	MoonshotV1_8KVision = "moonshot-v1-8k-vision-preview"

	// MoonshotV1_32KVision understands image content and outputs text, 32k context.
	MoonshotV1_32KVision = "moonshot-v1-32k-vision-preview"

	// MoonshotV1_128KVision understands image content and outputs text, 128k context.
	MoonshotV1_128KVision = "moonshot-v1-128k-vision-preview"
)

const (
	// OllamaModelsURL is the official Ollama models library page.
	// Use this to check for new models and model updates.
	OllamaModelsURL = "https://ollama.com/library"

	// OllamaAPIURL is the Ollama API reference page.
	OllamaAPIURL = "https://github.com/ollama/ollama/blob/main/docs/api.md"
)

const (
	OllamaLlama3_8B  = "llama3:8b"  // Llama 3 8B
	OllamaLlama3_70B = "llama3:70b" // Llama 3 70B
)

const (
	OllamaMistral7B   = "mistral:7b"   // Mistral 7B
	OllamaMixtral8x7B = "mixtral:8x7b" // Mixtral 8x7B
)

const (
	OllamaCodeLlama = "codellama:13b"       // CodeLlama 13B
	OllamaDeepSeek  = "deepseek-coder:6.7b" // DeepSeek Coder 6.7B
)

const (
	OllamaGemma2B = "gemma:2b" // Gemma 2B
	OllamaGemma7B = "gemma:7b" // Gemma 7B
)

const (
	// OpenAIModelsURL is the official OpenAI models documentation page.
	// Use this to check for new models, deprecations, and model updates.
	OpenAIModelsURL = "https://platform.openai.com/docs/models"

	// OpenAIAPIURL is the OpenAI API reference page.
	OpenAIAPIURL = "https://platform.openai.com/docs/api-reference"
)

const (
	GPT5           = "gpt-5"             // GPT-5 latest
	GPT5Mini       = "gpt-5-mini"        // GPT-5 Mini
	GPT5Nano       = "gpt-5-nano"        // GPT-5 Nano
	GPT5ChatLatest = "gpt-5-chat-latest" // GPT-5 Chat Latest
)

const (
	GPT4_1     = "gpt-4.1"      // GPT-4.1
	GPT4_1Mini = "gpt-4.1-mini" // GPT-4.1 Mini
	GPT4_1Nano = "gpt-4.1-nano" // GPT-4.1 Nano
)

const (
	GPT4o     = "gpt-4o"      // GPT-4o
	GPT4oMini = "gpt-4o-mini" // GPT-4o Mini
)

const (
	// QwenModelsURL is the official Qwen model documentation page.
	// Use this to check for new models, deprecations, and model updates.
	QwenModelsURL = "https://help.aliyun.com/zh/model-studio/getting-started/models"

	// QwenAPIURL is the Qwen DashScope API reference page.
	QwenAPIURL = "https://help.aliyun.com/zh/model-studio/developer-reference/use-qwen-by-calling-api"
)

const (
	// Qwen3Max is the latest flagship reasoning model with thinking capability.
	// Strong instruction following and complex task handling.
	Qwen3Max = "qwen3-max"

	// Qwen3MaxPreview is the preview version with extended thinking capabilities.
	Qwen3MaxPreview = "qwen3-max-preview"

	// QwenMax is the flagship reasoning model (International/China regions only).
	QwenMax = "qwen-max"
)

const (
	// Qwen3_5Plus is an advanced balanced model with thinking mode support.
	Qwen3_5Plus = "qwen3.5-plus"

	// QwenPlus is the balanced performance model with thinking mode.
	// Wide availability including US region.
	QwenPlus = "qwen-plus"

	// Qwen3_5Flash is an ultra-fast lightweight model for high-throughput.
	Qwen3_5Flash = "qwen3.5-flash"

	// QwenFlash is the fast lightweight model with context caching.
	// Wide availability including US region.
	QwenFlash = "qwen-flash"

	// QwenTurbo is the fastest lightweight model (deprecated, use QwenFlash).
	// Deprecated: use QwenFlash instead.
	QwenTurbo = "qwen-turbo"
)

const (
	// QwQPlus is the deep reasoning model with extended chain-of-thought.
	QwQPlus = "qwq-plus"

	// QwQ32B is the open-source 32B reasoning model with powerful logic capabilities.
	QwQ32B = "qwq-32b"
)

const (
	// Qwen3_5_397B is the largest open-source model with 397B parameters.
	Qwen3_5_397B = "qwen3.5-397b-a17b"

	// Qwen3_5_122B is the large open-source model with 122B parameters.
	Qwen3_5_122B = "qwen3.5-122b-a10b"

	// Qwen3_5_27B is the medium open-source model with 27B parameters.
	Qwen3_5_27B = "qwen3.5-27b"

	// Qwen3_5_35B is the efficient open-source model with 35B parameters.
	Qwen3_5_35B = "qwen3.5-35b-a3b"
)

const (
	// Qwen3_235B is the dual-mode 235B model supporting thinking and non-thinking.
	Qwen3_235B = "qwen3-235b-a22b"

	// Qwen3_32B is the versatile 32B model with dual-mode capabilities.
	Qwen3_32B = "qwen3-32b"

	// Qwen3_30B is the efficient 30B model with MoE architecture.
	Qwen3_30B = "qwen3-30b-a3b"

	// Qwen3_14B is the medium-sized 14B model with good performance-cost balance.
	Qwen3_14B = "qwen3-14b"

	// Qwen3_8B is the compact 8B model optimized for efficiency.
	Qwen3_8B = "qwen3-8b"
)

const (
	// Qwen2_5_72B is the large 72B instruction-following model.
	Qwen2_5_72B = "qwen2.5-72b-instruct"

	// Qwen2_5_32B is the medium 32B instruction-following model.
	Qwen2_5_32B = "qwen2.5-32b-instruct"

	// Qwen2_5_14B is the compact 14B instruction-following model.
	Qwen2_5_14B = "qwen2.5-14b-instruct"

	// Qwen2_5_7B is the small 7B instruction-following model.
	Qwen2_5_7B = "qwen2.5-7b-instruct"

	// Qwen2_5_14B_1M is the extended context 14B model supporting up to 1M tokens.
	Qwen2_5_14B_1M = "qwen2.5-14b-instruct-1m"

	// Qwen2_5_7B_1M is the extended context 7B model supporting up to 1M tokens.
	Qwen2_5_7B_1M = "qwen2.5-7b-instruct-1m"
)

const (
	// VertexModelsURL is the official Google Vertex AI models documentation page.
	// Use this to check for new models, deprecations, and model updates.
	VertexModelsURL = "https://cloud.google.com/vertex-ai/generative-ai/docs/learn/models"

	// VertexAPIURL is the Google Vertex AI API reference page.
	VertexAPIURL = "https://cloud.google.com/vertex-ai/generative-ai/docs/reference/rest"
)

const (
	// XAIModelsURL is the official X.AI models documentation page.
	// Use this to check for new models, deprecations, and model updates.
	XAIModelsURL = "https://docs.x.ai/docs/models"

	// XAIapiurl is the X.AI API reference page.
	XAIAPIURL = "https://docs.x.ai/docs"
)

const (
	// Grok4_1FastReasoning is the best tool-calling model with 2M context window.
	// Optimized for maximum intelligence and agentic tool calling.
	Grok4_1FastReasoning = "grok-4-1-fast-reasoning"

	// Grok4_1FastNonReasoning provides instant responses with 2M context window.
	// Optimized for speed without reasoning overhead.
	Grok4_1FastNonReasoning = "grok-4-1-fast-non-reasoning"
)

const (
	// Grok4_0709 is the flagship Grok 4 model with 256K context window.
	// Released July 9, 2025. Provides high-quality reasoning.
	Grok4_0709 = "grok-4-0709"

	// Grok4FastReasoning provides fast reasoning with 2M context window.
	// Performance on par with grok-4-0709 with larger context.
	Grok4FastReasoning = "grok-4-fast-reasoning"

	// Grok4FastNonReasoning provides fast non-reasoning with 2M context window.
	// Optimized for speed without reasoning overhead.
	Grok4FastNonReasoning = "grok-4-fast-non-reasoning"

	// GrokCodeFast1 is optimized for agentic coding with 256K context window.
	// Speedy and economical reasoning model for coding tasks.
	GrokCodeFast1 = "grok-code-fast-1"
)

const (
	Grok3     = "grok-3"      // Grok 3
	Grok3Mini = "grok-3-mini" // Grok 3 Mini (smaller, faster)
)

const (
	Grok2_1212   = "grok-2-1212"        // Grok 2 (December 2024)
	Grok2_Vision = "grok-2-vision-1212" // Grok 2 with vision capabilities
)

const (
	// GrokBeta is deprecated. Use Grok3 or Grok4_1FastReasoning instead.
	GrokBeta = "grok-beta" // Deprecated: use grok-3 or grok-4

	// GrokVision is deprecated. Use Grok2_Vision instead.
	GrokVision = "grok-vision-beta" // Deprecated
)

const (
	// BedrockTitan is Amazon Titan Text Express on AWS Bedrock.
	BedrockTitan = "amazon.titan-text-express-v1"
)

const (
	Claude3_5Haiku = "claude-3-5-haiku-20241022" // Claude 3.5 Haiku (October 2024)
)

const (
	Claude3_7Sonnet = "claude-3-7-sonnet-20250219" // Claude 3.7 Sonnet (February 2025)
)

const (
	ClaudeSonnet4 = "claude-sonnet-4-20250514" // Claude Sonnet 4 (May 2025)
)

const (
	// GLM4_32B is an ultra-budget dense 32B model, 128K context, no thinking mode.
	GLM4_32B = "glm-4-32b-0414-128k"
)

const (
	// GLM4_6 is a balanced model with 200K context and auto-thinking.
	GLM4_6 = "glm-4.6"
)

const (
	GPT35Turbo = "gpt-3.5-turbo" // GPT-3.5 Turbo
)

const (
	GPT4Turbo = "gpt-4-turbo" // GPT-4 Turbo
)

const (
	GeminiPro = "gemini-pro" // Legacy Gemini Pro (use Gemini 2.5 instead)
)

const (
	// KimiK2_5 is the most intelligent model with native multimodal architecture.
	// Supports visual/text input, thinking/non-thinking modes, 256k context.
	KimiK2_5 = "kimi-k2.5"
)

const (
	OllamaQwen2_5 = "qwen2.5:7b" // Qwen 2.5 7B
)

const (
	// VertexClaudeOpus4 is Claude Opus 4 on Google Vertex AI.
	VertexClaudeOpus4 = "claude-opus-4@20250514"
)