@ailib-official/ai-protocol 0.8.4

package/v1/models/ernie.yaml

@@ -0,0 +1,146 @@
+$schema: "https://raw.githubusercontent.com/ailib-official/ai-protocol/main/schemas/v1.json"
+protocol_version: "1.1"
+
+# 百度文心 (Baidu ERNIE) Model Instances
+#
+# VERIFIED_AT: 2026-01-31 (based on public Baidu Cloud documentation)
+#
+# Notes:
+# - ERNIE (文心一言) is Baidu's flagship LLM series.
+# - Different models have different API endpoints.
+
+models:
+  # ERNIE 4.0 series (flagship)
+  ernie-4.0-8k:
+    provider: baidu
+    model_id: "ernie-4.0-8k"
+    display_name: "文心一言 4.0 8K"
+    verification:
+      status: unverified
+      verified_at: "2026-01-31"
+      source: "documentation"
+      notes: "ERNIE 4.0 flagship model with enhanced reasoning."
+    context_window: 8192
+    capabilities: [chat, tools, streaming]
+    status: active
+    tags: ["baidu", "ernie", "flagship", "cn"]
+    pricing:
+      input_per_token: 0.00012
+      output_per_token: 0.00012
+
+  ernie-4.0-turbo-8k:
+    provider: baidu
+    model_id: "ernie-4.0-turbo-8k"
+    display_name: "文心一言 4.0 Turbo 8K"
+    verification:
+      status: unverified
+      verified_at: "2026-01-31"
+      source: "documentation"
+    context_window: 8192
+    capabilities: [chat, tools, streaming]
+    status: active
+    tags: ["baidu", "ernie", "turbo", "cn"]
+    pricing:
+      input_per_token: 0.00003
+      output_per_token: 0.00006
+
+  # ERNIE 3.5 series
+  ernie-3.5-8k:
+    provider: baidu
+    model_id: "ernie-3.5-8k"
+    display_name: "文心一言 3.5 8K"
+    verification:
+      status: unverified
+      verified_at: "2026-01-31"
+      source: "documentation"
+    context_window: 8192
+    capabilities: [chat, tools, streaming]
+    status: active
+    tags: ["baidu", "ernie", "cn"]
+    pricing:
+      input_per_token: 0.000012
+      output_per_token: 0.000012
+
+  ernie-3.5-128k:
+    provider: baidu
+    model_id: "ernie-3.5-128k"
+    display_name: "文心一言 3.5 128K"
+    verification:
+      status: unverified
+      verified_at: "2026-01-31"
+      source: "documentation"
+    context_window: 131072
+    capabilities: [chat, tools, streaming]
+    status: active
+    tags: ["baidu", "ernie", "long_context", "cn"]
+    pricing:
+      input_per_token: 0.000024
+      output_per_token: 0.000024
+
+  # ERNIE Speed series (cost-effective)
+  ernie-speed-8k:
+    provider: baidu
+    model_id: "ernie-speed-8k"
+    display_name: "文心一言 Speed 8K"
+    verification:
+      status: unverified
+      verified_at: "2026-01-31"
+      source: "documentation"
+    context_window: 8192
+    capabilities: [chat, streaming]
+    status: active
+    tags: ["baidu", "ernie", "speed", "cost_effective", "cn"]
+    pricing:
+      input_per_token: 0.000001
+      output_per_token: 0.000001
+
+  ernie-speed-128k:
+    provider: baidu
+    model_id: "ernie-speed-128k"
+    display_name: "文心一言 Speed 128K"
+    verification:
+      status: unverified
+      verified_at: "2026-01-31"
+      source: "documentation"
+    context_window: 131072
+    capabilities: [chat, streaming]
+    status: active
+    tags: ["baidu", "ernie", "speed", "long_context", "cn"]
+    pricing:
+      input_per_token: 0.000001
+      output_per_token: 0.000001
+
+  # ERNIE Lite series (free tier available)
+  ernie-lite-8k:
+    provider: baidu
+    model_id: "ernie-lite-8k"
+    display_name: "文心一言 Lite 8K"
+    verification:
+      status: unverified
+      verified_at: "2026-01-31"
+      source: "documentation"
+      notes: "Free tier available with rate limits."
+    context_window: 8192
+    capabilities: [chat, streaming]
+    status: active
+    tags: ["baidu", "ernie", "lite", "free_tier", "cn"]
+    pricing:
+      input_per_token: 0
+      output_per_token: 0
+
+  # ERNIE Character (角色扮演)
+  ernie-character-8k:
+    provider: baidu
+    model_id: "ernie-char-8k"
+    display_name: "文心一言 Character 8K"
+    verification:
+      status: unverified
+      verified_at: "2026-01-31"
+      source: "documentation"
+    context_window: 8192
+    capabilities: [chat, streaming]
+    status: active
+    tags: ["baidu", "ernie", "character", "roleplay", "cn"]
+    pricing:
+      input_per_token: 0.000004
+      output_per_token: 0.000008

package/v1/models/fireworks.yaml

@@ -0,0 +1,145 @@
+$schema: "https://raw.githubusercontent.com/ailib-official/ai-protocol/main/schemas/v1.json"
+protocol_version: "1.1"
+
+# Fireworks AI Model Instances
+#
+# VERIFIED_AT: 2026-01-29 (based on public Fireworks AI documentation)
+#
+# Notes:
+# - Fireworks AI provides high-performance inference for various open-source models.
+# - Model IDs follow the pattern: accounts/fireworks/models/{model_name}
+
+models:
+  # Llama 3.1 models
+  "accounts/fireworks/models/llama-v3p1-8b-instruct":
+    provider: fireworks
+    model_id: "accounts/fireworks/models/llama-v3p1-8b-instruct"
+    display_name: "Llama 3.1 8B Instruct"
+    verification:
+      status: unverified
+      verified_at: "2026-01-29"
+      source: "documentation"
+      notes: "Based on public documentation. Runtime verification recommended."
+    context_window: 131072
+    capabilities: [chat, tools, streaming]
+    status: active
+    tags: ["meta", "llama", "fireworks", "instruct"]
+    pricing:
+      input_per_token: 0.0000002
+      output_per_token: 0.0000002
+
+  "accounts/fireworks/models/llama-v3p1-70b-instruct":
+    provider: fireworks
+    model_id: "accounts/fireworks/models/llama-v3p1-70b-instruct"
+    display_name: "Llama 3.1 70B Instruct"
+    verification:
+      status: unverified
+      verified_at: "2026-01-29"
+      source: "documentation"
+    context_window: 131072
+    capabilities: [chat, tools, streaming]
+    status: active
+    tags: ["meta", "llama", "fireworks", "instruct"]
+    pricing:
+      input_per_token: 0.0000009
+      output_per_token: 0.0000009
+
+  "accounts/fireworks/models/llama-v3p1-405b-instruct":
+    provider: fireworks
+    model_id: "accounts/fireworks/models/llama-v3p1-405b-instruct"
+    display_name: "Llama 3.1 405B Instruct"
+    verification:
+      status: unverified
+      verified_at: "2026-01-29"
+      source: "documentation"
+    context_window: 131072
+    capabilities: [chat, tools, streaming]
+    status: active
+    tags: ["meta", "llama", "fireworks", "instruct", "flagship"]
+    pricing:
+      input_per_token: 0.000003
+      output_per_token: 0.000003
+
+  # Mixtral models
+  "accounts/fireworks/models/mixtral-8x7b-instruct":
+    provider: fireworks
+    model_id: "accounts/fireworks/models/mixtral-8x7b-instruct"
+    display_name: "Mixtral 8x7B Instruct"
+    verification:
+      status: unverified
+      verified_at: "2026-01-29"
+      source: "documentation"
+    context_window: 32768
+    capabilities: [chat, tools, streaming]
+    status: active
+    tags: ["mistral", "mixtral", "moe", "fireworks"]
+    pricing:
+      input_per_token: 0.0000005
+      output_per_token: 0.0000005
+
+  "accounts/fireworks/models/mixtral-8x22b-instruct":
+    provider: fireworks
+    model_id: "accounts/fireworks/models/mixtral-8x22b-instruct"
+    display_name: "Mixtral 8x22B Instruct"
+    verification:
+      status: unverified
+      verified_at: "2026-01-29"
+      source: "documentation"
+    context_window: 65536
+    capabilities: [chat, tools, streaming]
+    status: active
+    tags: ["mistral", "mixtral", "moe", "fireworks"]
+    pricing:
+      input_per_token: 0.0000012
+      output_per_token: 0.0000012
+
+  # Qwen models
+  "accounts/fireworks/models/qwen2p5-72b-instruct":
+    provider: fireworks
+    model_id: "accounts/fireworks/models/qwen2p5-72b-instruct"
+    display_name: "Qwen 2.5 72B Instruct"
+    verification:
+      status: unverified
+      verified_at: "2026-01-29"
+      source: "documentation"
+    context_window: 32768
+    capabilities: [chat, tools, streaming]
+    status: active
+    tags: ["alibaba", "qwen", "fireworks"]
+    pricing:
+      input_per_token: 0.0000009
+      output_per_token: 0.0000009
+
+  # DeepSeek models
+  "accounts/fireworks/models/deepseek-v3":
+    provider: fireworks
+    model_id: "accounts/fireworks/models/deepseek-v3"
+    display_name: "DeepSeek V3"
+    verification:
+      status: unverified
+      verified_at: "2026-01-29"
+      source: "documentation"
+    context_window: 131072
+    capabilities: [chat, tools, streaming]
+    status: active
+    tags: ["deepseek", "fireworks", "moe"]
+    pricing:
+      input_per_token: 0.0000009
+      output_per_token: 0.0000009
+
+  # FireFunction (fine-tuned for function calling)
+  "accounts/fireworks/models/firefunction-v2":
+    provider: fireworks
+    model_id: "accounts/fireworks/models/firefunction-v2"
+    display_name: "FireFunction V2"
+    verification:
+      status: unverified
+      verified_at: "2026-01-29"
+      source: "documentation"
+    context_window: 8192
+    capabilities: [chat, tools, streaming, function_calling]
+    status: active
+    tags: ["fireworks", "function_calling", "specialized"]
+    pricing:
+      input_per_token: 0.0000009
+      output_per_token: 0.0000009

package/v1/models/gemini.yaml

@@ -0,0 +1,32 @@
+$schema: "https://raw.githubusercontent.com/ailib-official/ai-protocol/main/schemas/v1.json"
+protocol_version: "1.1"
+
+# Gemini Model Instances
+models:
+  # VERIFIED_AT: 2026-01-28 (via Gemini list_models: GET /v1beta/models)
+  gemini-2.5-pro:
+    provider: gemini
+    model_id: "gemini-2.5-pro"
+    name: "Gemini 2.5 Pro"
+    verification:
+      status: verified
+      verified_at: "2026-01-28"
+      source: "list_models:/v1beta/models"
+    context_window: 2000000
+    capabilities: ["chat", "vision", "audio", "video", "tools", "streaming", "reasoning"]
+    status: active
+    tags: ["google", "multimodal", "large-context"]
+
+  # VERIFIED_AT: 2026-01-28 (via Gemini list_models: GET /v1beta/models)
+  gemini-2.5-flash:
+    provider: gemini
+    model_id: "gemini-2.5-flash"
+    name: "Gemini 2.5 Flash"
+    verification:
+      status: verified
+      verified_at: "2026-01-28"
+      source: "list_models:/v1beta/models"
+    context_window: 1000000
+    capabilities: ["chat", "vision", "audio", "video", "tools", "streaming"]
+    status: active
+    tags: ["google", "multimodal", "fast", "cost-effective"]

package/v1/models/gpt.yaml

@@ -0,0 +1,90 @@
+$schema: "https://raw.githubusercontent.com/ailib-official/ai-protocol/main/schemas/v1.json"
+protocol_version: "1.1"
+
+# GPT Model Instances
+models:
+  gpt-4o:
+    provider: openai
+    model_id: "gpt-4o"
+    display_name: "GPT-4o"
+    verification:
+      status: verified
+      verified_at: "2026-01-28"
+      source: "list_models:/models"
+    context_window: 128000
+    capabilities: [chat, vision, tools, streaming, agentic, parallel_tools, reasoning]
+    pricing:
+      input_per_token: 0.000005
+      output_per_token: 0.000015
+    status: active
+    tags: ["gpt", "vision", "tools", "agentic", "reasoning"]
+    agentic_capabilities:
+      reasoning_effort: "high"
+      thinking_blocks: false
+      parallel_tools: true
+      max_parallel_tools: 5
+      builtin_tools: ["web_search", "code_execution"]
+
+  gpt-4o-mini:
+    provider: openai
+    model_id: "gpt-4o-mini"
+    display_name: "GPT-4o Mini"
+    verification:
+      status: verified
+      verified_at: "2026-01-28"
+      source: "list_models:/models"
+    context_window: 128000
+    capabilities: [chat, vision, tools, streaming, agentic, parallel_tools]
+    pricing:
+      input_per_token: 0.00000015
+      output_per_token: 0.0000006
+    status: active
+    tags: ["gpt", "vision", "tools", "cost-effective", "agentic"]
+
+  gpt-4-turbo:
+    provider: openai
+    model_id: "gpt-4-turbo"
+    display_name: "GPT-4 Turbo"
+    verification:
+      status: verified
+      verified_at: "2026-01-28"
+      source: "list_models:/models"
+      notes: "Not present in current /models list for this key; marked deprecated."
+    context_window: 128000
+    capabilities: [chat, vision, tools, streaming, agentic, parallel_tools]
+    pricing:
+      input_per_token: 0.00001
+      output_per_token: 0.00003
+    # VERIFIED_AT: 2026-01-28 (via OpenAI list_models: GET /v1/models) -> not present for this key.
+    status: deprecated
+    tags: ["gpt", "vision", "tools", "agentic"]
+
+  # VERIFIED_AT: 2026-01-28 (via OpenAI list_models: GET /v1/models)
+  gpt-4.1:
+    provider: openai
+    model_id: "gpt-4.1"
+    display_name: "GPT-4.1"
+    verification:
+      status: verified
+      verified_at: "2026-01-28"
+      source: "list_models:/models"
+    context_window: 128000
+    capabilities: [chat, vision, tools, streaming, agentic, parallel_tools, reasoning]
+    status: active
+    tags: ["gpt", "vision", "tools", "agentic", "reasoning"]
+
+  gpt-35-turbo:
+    provider: openai
+    model_id: "gpt-3.5-turbo"
+    display_name: "GPT-3.5 Turbo"
+    verification:
+      status: verified
+      verified_at: "2026-01-28"
+      source: "list_models:/models"
+    context_window: 16385
+    capabilities: [chat, tools, streaming, agentic]
+    pricing:
+      input_per_token: 0.0000015
+      output_per_token: 0.000002
+    status: active
+    tags: ["gpt", "tools", "cost-effective", "agentic"]

package/v1/models/grok.yaml

@@ -0,0 +1,74 @@
+$schema: "https://raw.githubusercontent.com/ailib-official/ai-protocol/main/schemas/v1.json"
+protocol_version: "1.5"
+
+# Grok Model Instances (xAI)
+models:
+  grok-2:
+    provider: xai
+    model_id: "grok-2"
+    display_name: "Grok-2"
+    verification:
+      status: unverified
+      source: "manual"
+      notes: "Pending API access for verification"
+    context_window: 128000
+    capabilities: [chat, vision, tools, streaming, agentic, parallel_tools, reasoning]
+    pricing:
+      input_per_token: null
+      output_per_token: null
+    status: active
+    tags: ["grok", "vision", "reasoning", "agentic", "real-time"]
+    agentic_capabilities:
+      reasoning_effort: "high"
+      thinking_blocks: false
+      parallel_tools: true
+      max_parallel_tools: 5
+      builtin_tools: ["web_search", "knowledge_access"]
+
+  grok-1.5-vision:
+    provider: xai
+    model_id: "grok-1.5-vision"
+    display_name: "Grok-1.5 Vision"
+    verification:
+      status: unverified
+      source: "manual"
+      notes: "Multimodal version with vision capabilities"
+    context_window: 128000
+    capabilities: [chat, vision, tools, streaming, agentic, parallel_tools, reasoning]
+    pricing:
+      input_per_token: null
+      output_per_token: null
+    status: active
+    tags: ["grok", "vision", "multimodal", "agentic", "reasoning"]
+
+  grok-1.5:
+    provider: xai
+    model_id: "grok-1.5"
+    display_name: "Grok-1.5"
+    verification:
+      status: unverified
+      source: "manual"
+      notes: "Enhanced reasoning capabilities"
+    context_window: 128000
+    capabilities: [chat, vision, tools, streaming, agentic, parallel_tools, reasoning]
+    pricing:
+      input_per_token: null
+      output_per_token: null
+    status: active
+    tags: ["grok", "reasoning", "agentic", "streaming"]
+
+  grok-beta:
+    provider: xai
+    model_id: "grok-beta"
+    display_name: "Grok Beta"
+    verification:
+      status: unverified
+      source: "manual"
+      notes: "Early access version"
+    context_window: 48000
+    capabilities: [chat, tools, streaming, agentic]
+    pricing:
+      input_per_token: null
+      output_per_token: null
+    status: beta
+    tags: ["grok", "beta", "agentic"]

package/v1/models/huggingface.yaml

@@ -0,0 +1,110 @@
+$schema: "https://raw.githubusercontent.com/ailib-official/ai-protocol/main/schemas/v1.json"
+protocol_version: "1.5"
+
+# Hugging Face Popular Model Instances
+# Notes: Hugging Face has 100K+ models; this is a curated list of popular open-source models
+# Model ID format: {organization}/{model-name}
+# Use connection_vars or path parameter to specify dynamic model_id
+
+models:
+  # Meta Llama Series
+  llama-3.1-70b-instruct:
+    provider: huggingface
+    model_id: "meta-llama/Meta-Llama-3.1-70B-Instruct"
+    display_name: "Llama 3.1 70B Instruct"
+    verification:
+      status: verified
+      verified_at: "2026-01-31"
+      source: "models_hub"
+      notes: "Popular open-source model on Hugging Face"
+    context_window: 131072
+    capabilities: [chat, tools, streaming, agentic]
+    pricing:
+      input_per_token: null
+      output_per_token: null
+    status: active
+    tags: ["llama", "meta", "open-source", "instruct", "large-context"]
+
+  llama-3.1-8b-instruct:
+    provider: huggingface
+    model_id: "meta-llama/Meta-Llama-3.1-8B-Instruct"
+    display_name: "Llama 3.1 8B Instruct"
+    verification:
+      status: verified
+      verified_at: "2026-01-31"
+      source: "models_hub"
+    context_window: 131072
+    capabilities: [chat, tools, streaming, agentic]
+    pricing:
+      input_per_token: null
+      output_per_token: null
+    status: active
+    tags: ["llama", "meta", "open-source", "fast", "cost-effective"]
+
+  # Mistral Series
+  mistral-large:
+    provider: huggingface
+    model_id: "mistralai/Mistral-Large-Instruct-2407"
+    display_name: "Mistral Large"
+    verification:
+      status: verified
+      verified_at: "2026-01-31"
+      source: "models_hub"
+    context_window: 128000
+    capabilities: [chat, tools, streaming, agentic, reasoning]
+    pricing:
+      input_per_token: null
+      output_per_token: null
+    status: active
+    tags: ["mistral", "open-source", "reasoning", "agentic"]
+
+  # Qwen Series
+  qwen-2.5-72b-instruct:
+    provider: huggingface
+    model_id: "Qwen/Qwen2.5-72B-Instruct"
+    display_name: "Qwen 2.5 72B Instruct"
+    verification:
+      status: verified
+      verified_at: "2026-01-31"
+      source: "models_hub"
+    context_window: 131072
+    capabilities: [chat, vision, tools, streaming, agentic, reasoning]
+    pricing:
+      input_per_token: null
+      output_per_token: null
+    status: active
+    tags: ["qwen", "alibaba", "open-source", "multimodal", "reasoning"]
+
+  # DeepSeek Series
+  deepseek-coder:
+    provider: huggingface
+    model_id: "deepseek-ai/DeepSeek-Coder-V2-Lite-Instruct"
+    display_name: "DeepSeek Coder V2 Lite"
+    verification:
+      status: verified
+      verified_at: "2026-01-31"
+      source: "models_hub"
+    context_window: 128000
+    capabilities: [chat, tools, streaming, agentic, reasoning]
+    pricing:
+      input_per_token: null
+      output_per_token: null
+    status: active
+    tags: ["deepseek", "code", "open-source", "reasoning"]
+
+  # Code Models
+  codestral:
+    provider: huggingface
+    model_id: "mistralai/Codestral-22B-v0.1"
+    display_name: "Codestral"
+    verification:
+      status: verified
+      verified_at: "2026-01-31"
+      source: "models_hub"
+    context_window: 32000
+    capabilities: [chat, tools, streaming, agentic, reasoning]
+    pricing:
+      input_per_token: null
+      output_per_token: null
+    status: active
+    tags: ["mistral", "code", "open-source", "agentic"]