@ailib-official/ai-protocol 0.8.4

package/v1/models/mistral.yaml

@@ -0,0 +1,100 @@
+$schema: "https://raw.githubusercontent.com/ailib-official/ai-protocol/main/schemas/v1.json"
+protocol_version: "1.1"
+
+# Mistral Model Instances
+# Reference: https://docs.mistral.ai/api/
+
+models:
+  mistral-small-latest:
+    provider: mistral
+    model_id: "mistral-small-latest"
+    display_name: "Mistral Small"
+    verification:
+      status: verified
+      verified_at: "2026-01-28"
+      source: "list_models:/models"
+    context_window: 32000
+    capabilities: [chat, tools, streaming, parallel_tools]
+    status: active
+    tags: ["mistral", "small", "tools", "latest"]
+
+  mistral-medium-latest:
+    provider: mistral
+    model_id: "mistral-medium-latest"
+    display_name: "Mistral Medium"
+    verification:
+      status: verified
+      verified_at: "2026-01-28"
+      source: "list_models:/models"
+    context_window: 32000
+    capabilities: [chat, tools, streaming, parallel_tools]
+    status: active
+    tags: ["mistral", "medium", "tools", "latest"]
+
+  mistral-large-latest:
+    provider: mistral
+    model_id: "mistral-large-latest"
+    display_name: "Mistral Large"
+    verification:
+      status: verified
+      verified_at: "2026-01-28"
+      source: "list_models:/models"
+    context_window: 32000
+    capabilities: [chat, tools, streaming, parallel_tools, agentic]
+    status: active
+    tags: ["mistral", "large", "tools", "agentic", "latest"]
+
+  pixtral-12b-latest:
+    provider: mistral
+    model_id: "pixtral-12b-latest"
+    display_name: "Pixtral 12B"
+    verification:
+      status: verified
+      verified_at: "2026-01-28"
+      source: "list_models:/models"
+    context_window: 128000
+    capabilities: [chat, vision, tools, streaming, parallel_tools]
+    status: active
+    tags: ["mistral", "pixtral", "vision", "12b", "latest"]
+
+  pixtral-large-latest:
+    provider: mistral
+    model_id: "pixtral-large-latest"
+    display_name: "Pixtral Large"
+    verification:
+      status: verified
+      verified_at: "2026-01-28"
+      source: "list_models:/models"
+    context_window: 128000
+    capabilities: [chat, vision, tools, streaming, parallel_tools, agentic]
+    status: active
+    tags: ["mistral", "pixtral", "vision", "large", "agentic", "latest"]
+
+  open-mistral-7b:
+    provider: mistral
+    model_id: "open-mistral-7b"
+    display_name: "Open Mistral 7B"
+    verification:
+      status: verified
+      verified_at: "2026-01-28"
+      source: "list_models:/models"
+    context_window: 32000
+    capabilities: [chat, streaming]
+    status: active
+    tags: ["mistral", "open", "7b", "base"]
+
+  # NOTE: `open-mixtral-8x7b` is not present in the current /v1/models list for this provider.
+  # Keep it removed to prevent 404 model_not_found. If it returns, re-add with evidence.
+
+  mistral-tiny:
+    provider: mistral
+    model_id: "mistral-tiny"
+    display_name: "Mistral Tiny"
+    verification:
+      status: verified
+      verified_at: "2026-01-28"
+      source: "list_models:/models"
+    context_window: 32000
+    capabilities: [chat, streaming]
+    status: active
+    tags: ["mistral", "tiny", "cost-effective"]

package/v1/models/moonshot.yaml

@@ -0,0 +1,97 @@
+$schema: "https://raw.githubusercontent.com/ailib-official/ai-protocol/main/schemas/v1.json"
+protocol_version: "1.1"
+
+# 月之暗面 Kimi (Moonshot) Model Instances
+#
+# VERIFIED_AT: 2026-01-31 (based on public Moonshot documentation)
+#
+# Notes:
+# - Kimi is known for ultra-long context windows.
+# - Supports file understanding and web search.
+
+models:
+  # Moonshot v1 series
+  moonshot-v1-8k:
+    provider: moonshot
+    model_id: "moonshot-v1-8k"
+    display_name: "Kimi v1 8K"
+    verification:
+      status: unverified
+      verified_at: "2026-01-31"
+      source: "documentation"
+      notes: "Standard context model, cost-effective."
+    context_window: 8192
+    capabilities: [chat, tools, streaming]
+    status: active
+    tags: ["moonshot", "kimi", "cn"]
+    pricing:
+      input_per_token: 0.000012
+      output_per_token: 0.000012
+
+  moonshot-v1-32k:
+    provider: moonshot
+    model_id: "moonshot-v1-32k"
+    display_name: "Kimi v1 32K"
+    verification:
+      status: unverified
+      verified_at: "2026-01-31"
+      source: "documentation"
+    context_window: 32768
+    capabilities: [chat, tools, streaming]
+    status: active
+    tags: ["moonshot", "kimi", "long_context", "cn"]
+    pricing:
+      input_per_token: 0.000024
+      output_per_token: 0.000024
+
+  moonshot-v1-128k:
+    provider: moonshot
+    model_id: "moonshot-v1-128k"
+    display_name: "Kimi v1 128K"
+    verification:
+      status: unverified
+      verified_at: "2026-01-31"
+      source: "documentation"
+    context_window: 131072
+    capabilities: [chat, tools, streaming]
+    status: active
+    tags: ["moonshot", "kimi", "ultra_long_context", "cn"]
+    pricing:
+      input_per_token: 0.00006
+      output_per_token: 0.00006
+
+  # Kimi k1 series (latest with enhanced capabilities)
+  kimi-latest:
+    provider: moonshot
+    model_id: "kimi-latest"
+    display_name: "Kimi Latest"
+    verification:
+      status: unverified
+      verified_at: "2026-01-31"
+      source: "documentation"
+      notes: "Alias for the latest Kimi model."
+    context_window: 131072
+    capabilities: [chat, tools, vision, streaming]
+    status: active
+    tags: ["moonshot", "kimi", "latest", "cn"]
+    pricing:
+      input_per_token: 0.00006
+      output_per_token: 0.00006
+
+  # Moonshot web search model
+  moonshot-v1-auto:
+    provider: moonshot
+    model_id: "moonshot-v1-auto"
+    display_name: "Kimi v1 Auto"
+    verification:
+      status: unverified
+      verified_at: "2026-01-31"
+      source: "documentation"
+      notes: "Auto-selects context window based on input."
+    context_window: 131072
+    capabilities: [chat, tools, streaming]
+    status: active
+    tags: ["moonshot", "kimi", "auto", "cn"]
+    pricing:
+      input_per_token: 0.000024
+      output_per_token: 0.000024

package/v1/models/nvidia.yaml

@@ -0,0 +1,118 @@
+$schema: "https://raw.githubusercontent.com/ailib-official/ai-protocol/main/schemas/v1.json"
+protocol_version: "1.1"
+
+# NVIDIA API Catalog – top LLMs including Minimax and Kimi latest (verified via official API reference)
+# Source: https://docs.api.nvidia.com/nim/reference/llm-apis
+# Catalog: https://build.nvidia.com/explore/discover
+models:
+  nvidia-nemotron-4-340b-instruct:
+    provider: nvidia
+    model_id: "nvidia-nemotron-4-340b-instruct"
+    display_name: "Nemotron 4 340B Instruct"
+    verification:
+      status: verified
+      verified_at: "2026-02-04"
+      source: "https://docs.api.nvidia.com/nim/reference/llm-apis"
+      notes: "NVIDIA flagship 340B chat model; OpenAI-compatible chat completions."
+    context_window: 128000
+    capabilities: [chat, vision, tools, streaming, agentic, parallel_tools, reasoning]
+    status: active
+    tags: ["nvidia", "nemotron", "flagship", "reasoning", "tools"]
+
+  "nvidia/llama3-chatqa-1.5-70b":
+    provider: nvidia
+    model_id: "nvidia/llama3-chatqa-1.5-70b"
+    display_name: "Llama 3 ChatQA 1.5 70B"
+    verification:
+      status: verified
+      verified_at: "2026-02-04"
+      source: "https://docs.api.nvidia.com/nim/reference/llm-apis"
+      notes: "70B chat QA model on NVIDIA API Catalog."
+    context_window: 32768
+    capabilities: [chat, tools, streaming, agentic]
+    status: active
+    tags: ["nvidia", "llama", "chatqa", "70b"]
+
+  "deepseek-ai/deepseek-r1":
+    provider: nvidia
+    model_id: "deepseek-ai/deepseek-r1"
+    display_name: "DeepSeek R1 (via NVIDIA)"
+    verification:
+      status: verified
+      verified_at: "2026-02-04"
+      source: "https://docs.api.nvidia.com/nim/reference/llm-apis"
+      notes: "Reasoning model available on NVIDIA API Catalog."
+    context_window: 64000
+    capabilities: [chat, tools, streaming, reasoning]
+    status: active
+    tags: ["deepseek", "reasoning", "nvidia-catalog"]
+
+  "mistralai/mistral-large-2-instruct":
+    provider: nvidia
+    model_id: "mistralai/mistral-large-2-instruct"
+    display_name: "Mistral Large 2 Instruct (via NVIDIA)"
+    verification:
+      status: verified
+      verified_at: "2026-02-04"
+      source: "https://docs.api.nvidia.com/nim/reference/llm-apis"
+      notes: "Mistral flagship instruct model on NVIDIA API Catalog."
+    context_window: 128000
+    capabilities: [chat, vision, tools, streaming, agentic]
+    status: active
+    tags: ["mistral", "instruct", "nvidia-catalog"]
+
+  "mistralai/mixtral-8x22b-instruct":
+    provider: nvidia
+    model_id: "mistralai/mixtral-8x22b-instruct"
+    display_name: "Mixtral 8x22B Instruct (via NVIDIA)"
+    verification:
+      status: verified
+      verified_at: "2026-02-04"
+      source: "https://docs.api.nvidia.com/nim/reference/llm-apis"
+      notes: "MoE 8x22B instruct model on NVIDIA API Catalog."
+    context_window: 65536
+    capabilities: [chat, tools, streaming, agentic]
+    status: active
+    tags: ["mistral", "mixtral", "moe", "nvidia-catalog"]
+
+  "meta/llama3-70b":
+    provider: nvidia
+    model_id: "meta/llama3-70b"
+    display_name: "Llama 3 70B (via NVIDIA)"
+    verification:
+      status: verified
+      verified_at: "2026-02-04"
+      source: "https://docs.api.nvidia.com/nim/reference/llm-apis"
+      notes: "Meta Llama 3 70B on NVIDIA API Catalog."
+    context_window: 8192
+    capabilities: [chat, tools, streaming]
+    status: active
+    tags: ["meta", "llama", "70b", "nvidia-catalog"]
+
+  "moonshotai/kimi-k2-instruct":
+    provider: nvidia
+    model_id: "moonshotai/kimi-k2-instruct"
+    display_name: "Kimi K2 Instruct (via NVIDIA)"
+    verification:
+      status: verified
+      verified_at: "2026-02-04"
+      source: "https://docs.api.nvidia.com/nim/reference/moonshotai-kimi-k2-instruct"
+      notes: "Moonshot Kimi K2 MoE 1T/32B; agentic, tools, 128K context on NVIDIA API Catalog."
+    context_window: 128000
+    capabilities: [chat, tools, streaming, agentic, reasoning]
+    status: active
+    tags: ["moonshot", "kimi", "moe", "nvidia-catalog", "agentic"]
+
+  "minimaxai/minimax-m2":
+    provider: nvidia
+    model_id: "minimaxai/minimax-m2"
+    display_name: "MiniMax M2 (via NVIDIA)"
+    verification:
+      status: verified
+      verified_at: "2026-02-04"
+      source: "https://docs.api.nvidia.com/nim/reference/minimaxai-minimax-m2"
+      notes: "MiniMax M2 MoE 230B/10B; coding, reasoning, tool-use, 128K context on NVIDIA API Catalog."
+    context_window: 128000
+    capabilities: [chat, tools, streaming, agentic, reasoning]
+    status: active
+    tags: ["minimax", "moe", "nvidia-catalog", "coding", "agentic"]

package/v1/models/palmyra.yaml

@@ -0,0 +1,59 @@
+$schema: "https://raw.githubusercontent.com/ailib-official/ai-protocol/main/schemas/v1.json"
+protocol_version: "1.5"
+
+# Writer Model Instances (Palmyra Series)
+# Notes: Writer provides enterprise-focused AI writing and content generation
+
+models:
+  palmyra-x-004:
+    provider: writer
+    model_id: "palmyra-x-004"
+    display_name: "Palmyra X 004"
+    verification:
+      status: unverified
+      source: "manual"
+      notes: "Flagship model for enterprise writing"
+    context_window: 128000
+    capabilities: [chat, tools, streaming, agentic, reasoning]
+    pricing:
+      input_per_token: null
+      output_per_token: null
+    status: active
+    tags: ["palmyra", "writer", "enterprise", "writing", "reasoning"]
+    agentic_capabilities:
+      reasoning_effort: "medium"
+      thinking_blocks: false
+      parallel_tools: false
+      builtin_tools: ["web_search", "fact_check", "compliance"]
+
+  palmyra-med:
+    provider: writer
+    model_id: "palmyra-med"
+    display_name: "Palmyra Med"
+    verification:
+      status: unverified
+      source: "manual"
+      notes: "Specialized for medical and healthcare content"
+    context_window: 64000
+    capabilities: [chat, tools, streaming, agentic]
+    pricing:
+      input_per_token: null
+      output_per_token: null
+    status: active
+    tags: ["palmyra", "writer", "medical", "healthcare", "enterprise"]
+
+  palmyra-fin:
+    provider: writer
+    model_id: "palmyra-fin"
+    display_name: "Palmyra Fin"
+    verification:
+      status: unverified
+      source: "manual"
+      notes: "Specialized for financial content"
+    context_window: 64000
+    capabilities: [chat, tools, streaming, agentic]
+    pricing:
+      input_per_token: null
+      output_per_token: null
+    status: active
+    tags: ["palmyra", "writer", "finance", "enterprise"]

package/v1/models/perplexity.yaml

@@ -0,0 +1,97 @@
+$schema: "https://raw.githubusercontent.com/ailib-official/ai-protocol/main/schemas/v1.json"
+protocol_version: "1.1"
+
+# Perplexity Model Instances
+#
+# VERIFIED_AT: 2026-01-31 (based on public Perplexity documentation)
+#
+# Notes:
+# - Perplexity is known for search-augmented AI.
+# - "Online" models include real-time web search.
+# - "Chat" models are standard conversational models.
+
+models:
+  # Sonar Online models (with web search)
+  llama-3.1-sonar-small-128k-online:
+    provider: perplexity
+    model_id: "llama-3.1-sonar-small-128k-online"
+    display_name: "Sonar Small Online (8B)"
+    verification:
+      status: unverified
+      verified_at: "2026-01-31"
+      source: "documentation"
+      notes: "Llama 3.1 8B with real-time web search."
+    context_window: 127072
+    capabilities: [chat, streaming, online_search]
+    status: active
+    tags: ["perplexity", "sonar", "online", "search"]
+    pricing:
+      input_per_token: 0.0000002
+      output_per_token: 0.0000002
+
+  llama-3.1-sonar-large-128k-online:
+    provider: perplexity
+    model_id: "llama-3.1-sonar-large-128k-online"
+    display_name: "Sonar Large Online (70B)"
+    verification:
+      status: unverified
+      verified_at: "2026-01-31"
+      source: "documentation"
+      notes: "Llama 3.1 70B with real-time web search."
+    context_window: 127072
+    capabilities: [chat, streaming, online_search]
+    status: active
+    tags: ["perplexity", "sonar", "online", "search", "large"]
+    pricing:
+      input_per_token: 0.000001
+      output_per_token: 0.000001
+
+  llama-3.1-sonar-huge-128k-online:
+    provider: perplexity
+    model_id: "llama-3.1-sonar-huge-128k-online"
+    display_name: "Sonar Huge Online (405B)"
+    verification:
+      status: unverified
+      verified_at: "2026-01-31"
+      source: "documentation"
+      notes: "Llama 3.1 405B with real-time web search."
+    context_window: 127072
+    capabilities: [chat, streaming, online_search]
+    status: active
+    tags: ["perplexity", "sonar", "online", "search", "flagship"]
+    pricing:
+      input_per_token: 0.000005
+      output_per_token: 0.000005
+
+  # Sonar Chat models (without web search)
+  llama-3.1-sonar-small-128k-chat:
+    provider: perplexity
+    model_id: "llama-3.1-sonar-small-128k-chat"
+    display_name: "Sonar Small Chat (8B)"
+    verification:
+      status: unverified
+      verified_at: "2026-01-31"
+      source: "documentation"
+    context_window: 131072
+    capabilities: [chat, streaming]
+    status: active
+    tags: ["perplexity", "sonar", "chat"]
+    pricing:
+      input_per_token: 0.0000002
+      output_per_token: 0.0000002
+
+  llama-3.1-sonar-large-128k-chat:
+    provider: perplexity
+    model_id: "llama-3.1-sonar-large-128k-chat"
+    display_name: "Sonar Large Chat (70B)"
+    verification:
+      status: unverified
+      verified_at: "2026-01-31"
+      source: "documentation"
+    context_window: 131072
+    capabilities: [chat, streaming]
+    status: active
+    tags: ["perplexity", "sonar", "chat", "large"]
+    pricing:
+      input_per_token: 0.000001
+      output_per_token: 0.000001

package/v1/models/qwen.yaml

@@ -0,0 +1,27 @@
+$schema: "https://raw.githubusercontent.com/ailib-official/ai-protocol/main/schemas/v1.json"
+protocol_version: "1.1"
+
+# Qwen Model Instances
+models:
+  qwen-max:
+    provider: qwen
+    model_id: "qwen-max"
+    display_name: "Qwen Max"
+    verification:
+      status: unverified
+      source: "manual"
+    context_window: 262144
+    capabilities: [chat, tools, streaming]
+    status: active
+    tags: ["alibaba", "qwen", "strong"]
+
+  qwen-turbo:
+    provider: qwen
+    model_id: "qwen-turbo"
+    display_name: "Qwen Turbo"
+    verification:
+      status: unverified
+      source: "manual"
+    capabilities: [chat, vision, tools, streaming, agentic]
+    status: active
+    tags: ["alibaba", "qwen", "fast"]

package/v1/models/replicate.yaml

@@ -0,0 +1,136 @@
+$schema: "https://raw.githubusercontent.com/ailib-official/ai-protocol/main/schemas/v1.json"
+protocol_version: "1.1"
+
+# Replicate Model Instances
+#
+# VERIFIED_AT: 2026-01-29 (based on public Replicate documentation)
+#
+# Notes:
+# - Replicate is a model hosting platform with async prediction API.
+# - Model IDs follow the pattern: {owner}/{model_name}
+# - Supports both native Replicate API and OpenAI-compatible endpoints for some models.
+
+models:
+  # Meta Llama models
+  "meta/llama-2-70b-chat":
+    provider: replicate
+    model_id: "meta/llama-2-70b-chat"
+    display_name: "Llama 2 70B Chat"
+    verification:
+      status: unverified
+      verified_at: "2026-01-29"
+      source: "documentation"
+      notes: "Based on public documentation. Runtime verification recommended."
+    context_window: 4096
+    capabilities: [chat, streaming]
+    status: active
+    tags: ["meta", "llama", "replicate", "chat"]
+
+  "meta/meta-llama-3-70b-instruct":
+    provider: replicate
+    model_id: "meta/meta-llama-3-70b-instruct"
+    display_name: "Llama 3 70B Instruct"
+    verification:
+      status: unverified
+      verified_at: "2026-01-29"
+      source: "documentation"
+    context_window: 8192
+    capabilities: [chat, streaming]
+    status: active
+    tags: ["meta", "llama", "replicate", "instruct"]
+
+  "meta/meta-llama-3.1-405b-instruct":
+    provider: replicate
+    model_id: "meta/meta-llama-3.1-405b-instruct"
+    display_name: "Llama 3.1 405B Instruct"
+    verification:
+      status: unverified
+      verified_at: "2026-01-29"
+      source: "documentation"
+    context_window: 131072
+    capabilities: [chat, streaming]
+    status: active
+    tags: ["meta", "llama", "replicate", "instruct", "flagship"]
+
+  # Mistral models
+  "mistralai/mistral-7b-instruct-v0.2":
+    provider: replicate
+    model_id: "mistralai/mistral-7b-instruct-v0.2"
+    display_name: "Mistral 7B Instruct v0.2"
+    verification:
+      status: unverified
+      verified_at: "2026-01-29"
+      source: "documentation"
+    context_window: 8192
+    capabilities: [chat, streaming]
+    status: active
+    tags: ["mistral", "replicate", "instruct"]
+
+  "mistralai/mixtral-8x7b-instruct-v0.1":
+    provider: replicate
+    model_id: "mistralai/mixtral-8x7b-instruct-v0.1"
+    display_name: "Mixtral 8x7B Instruct"
+    verification:
+      status: unverified
+      verified_at: "2026-01-29"
+      source: "documentation"
+    context_window: 32768
+    capabilities: [chat, streaming]
+    status: active
+    tags: ["mistral", "mixtral", "moe", "replicate"]
+
+  # Stability AI models (text)
+  "stability-ai/stablelm-tuned-alpha-7b":
+    provider: replicate
+    model_id: "stability-ai/stablelm-tuned-alpha-7b"
+    display_name: "StableLM Tuned Alpha 7B"
+    verification:
+      status: unverified
+      verified_at: "2026-01-29"
+      source: "documentation"
+    context_window: 4096
+    capabilities: [chat, streaming]
+    status: deprecated
+    tags: ["stability", "stablelm", "replicate"]
+
+  # Snowflake models
+  "snowflake/snowflake-arctic-instruct":
+    provider: replicate
+    model_id: "snowflake/snowflake-arctic-instruct"
+    display_name: "Snowflake Arctic Instruct"
+    verification:
+      status: unverified
+      verified_at: "2026-01-29"
+      source: "documentation"
+    context_window: 4096
+    capabilities: [chat, streaming]
+    status: active
+    tags: ["snowflake", "arctic", "replicate", "enterprise"]
+
+  # Vision-language models
+  "yorickvp/llava-13b":
+    provider: replicate
+    model_id: "yorickvp/llava-13b"
+    display_name: "LLaVA 13B"
+    verification:
+      status: unverified
+      verified_at: "2026-01-29"
+      source: "documentation"
+    context_window: 2048
+    capabilities: [chat, vision, streaming]
+    status: active
+    tags: ["llava", "vision", "multimodal", "replicate"]
+
+  # Code models
+  "meta/codellama-70b-instruct":
+    provider: replicate
+    model_id: "meta/codellama-70b-instruct"
+    display_name: "Code Llama 70B Instruct"
+    verification:
+      status: unverified
+      verified_at: "2026-01-29"
+      source: "documentation"
+    context_window: 16384
+    capabilities: [chat, code, streaming]
+    status: active
+    tags: ["meta", "codellama", "code", "replicate"]