@ailib-official/ai-protocol 0.8.4

package/v1/models/baichuan.yaml

@@ -0,0 +1,95 @@
+$schema: "https://raw.githubusercontent.com/ailib-official/ai-protocol/main/schemas/v1.json"
+protocol_version: "1.1"
+
+# 百川智能 (Baichuan) Model Instances
+#
+# VERIFIED_AT: 2026-01-31 (based on public Baichuan documentation)
+#
+# Notes:
+# - Baichuan is a leading Chinese AI company.
+# - Known for strong Chinese language understanding.
+
+models:
+  # Baichuan 4 series (latest)
+  Baichuan4:
+    provider: baichuan
+    model_id: "Baichuan4"
+    display_name: "百川 4"
+    verification:
+      status: unverified
+      verified_at: "2026-01-31"
+      source: "documentation"
+      notes: "Latest flagship model with best performance."
+    context_window: 32768
+    capabilities: [chat, tools, streaming]
+    status: active
+    tags: ["baichuan", "flagship", "cn"]
+    pricing:
+      input_per_token: 0.0001
+      output_per_token: 0.0001
+
+  # Baichuan 3 Turbo series
+  Baichuan3-Turbo:
+    provider: baichuan
+    model_id: "Baichuan3-Turbo"
+    display_name: "百川 3 Turbo"
+    verification:
+      status: unverified
+      verified_at: "2026-01-31"
+      source: "documentation"
+    context_window: 32768
+    capabilities: [chat, tools, streaming]
+    status: active
+    tags: ["baichuan", "turbo", "cn"]
+    pricing:
+      input_per_token: 0.000012
+      output_per_token: 0.000012
+
+  Baichuan3-Turbo-128k:
+    provider: baichuan
+    model_id: "Baichuan3-Turbo-128k"
+    display_name: "百川 3 Turbo 128K"
+    verification:
+      status: unverified
+      verified_at: "2026-01-31"
+      source: "documentation"
+    context_window: 131072
+    capabilities: [chat, tools, streaming]
+    status: active
+    tags: ["baichuan", "turbo", "long_context", "cn"]
+    pricing:
+      input_per_token: 0.000024
+      output_per_token: 0.000024
+
+  # Baichuan 2 series (legacy)
+  Baichuan2-Turbo:
+    provider: baichuan
+    model_id: "Baichuan2-Turbo"
+    display_name: "百川 2 Turbo"
+    verification:
+      status: unverified
+      verified_at: "2026-01-31"
+      source: "documentation"
+    context_window: 8192
+    capabilities: [chat, streaming]
+    status: active
+    tags: ["baichuan", "turbo", "legacy", "cn"]
+    pricing:
+      input_per_token: 0.000008
+      output_per_token: 0.000008
+
+  Baichuan2-Turbo-192k:
+    provider: baichuan
+    model_id: "Baichuan2-Turbo-192k"
+    display_name: "百川 2 Turbo 192K"
+    verification:
+      status: unverified
+      verified_at: "2026-01-31"
+      source: "documentation"
+    context_window: 196608
+    capabilities: [chat, streaming]
+    status: active
+    tags: ["baichuan", "turbo", "ultra_long_context", "cn"]
+    pricing:
+      input_per_token: 0.000016
+      output_per_token: 0.000016

package/v1/models/cerebras.yaml

@@ -0,0 +1,99 @@
+$schema: "https://raw.githubusercontent.com/ailib-official/ai-protocol/main/schemas/v1.json"
+protocol_version: "1.1"
+
+# Cerebras Model Instances
+#
+# VERIFIED_AT: 2026-01-29 (based on public Cerebras documentation)
+#
+# Notes:
+# - Cerebras offers extremely fast inference using Wafer-Scale Engine technology.
+# - Known for achieving thousands of tokens per second.
+# - Primarily hosts Llama models.
+
+models:
+  # Llama 3.1 models
+  llama3.1-8b:
+    provider: cerebras
+    model_id: "llama3.1-8b"
+    display_name: "Llama 3.1 8B"
+    verification:
+      status: unverified
+      verified_at: "2026-01-29"
+      source: "documentation"
+      notes: "Based on public documentation. Cerebras achieves ~2100 tokens/sec."
+    context_window: 8192
+    capabilities: [chat, tools, streaming]
+    status: active
+    tags: ["meta", "llama", "cerebras", "fast"]
+    pricing:
+      input_per_token: 0.0000001
+      output_per_token: 0.0000001
+
+  llama3.1-70b:
+    provider: cerebras
+    model_id: "llama3.1-70b"
+    display_name: "Llama 3.1 70B"
+    verification:
+      status: unverified
+      verified_at: "2026-01-29"
+      source: "documentation"
+      notes: "Cerebras achieves ~450 tokens/sec for this model."
+    context_window: 8192
+    capabilities: [chat, tools, streaming]
+    status: active
+    tags: ["meta", "llama", "cerebras", "fast"]
+    pricing:
+      input_per_token: 0.0000006
+      output_per_token: 0.0000006
+
+  # Llama 3.3 models
+  llama-3.3-70b:
+    provider: cerebras
+    model_id: "llama-3.3-70b"
+    display_name: "Llama 3.3 70B"
+    verification:
+      status: unverified
+      verified_at: "2026-01-29"
+      source: "documentation"
+      notes: "Cerebras achieves ~970 tokens/sec for this model."
+    context_window: 8192
+    capabilities: [chat, tools, streaming]
+    status: active
+    tags: ["meta", "llama", "cerebras", "fast", "latest"]
+    pricing:
+      input_per_token: 0.00000085
+      output_per_token: 0.00000085
+
+  # Qwen models
+  qwen-2.5-32b:
+    provider: cerebras
+    model_id: "qwen-2.5-32b"
+    display_name: "Qwen 2.5 32B"
+    verification:
+      status: unverified
+      verified_at: "2026-01-29"
+      source: "documentation"
+    context_window: 8192
+    capabilities: [chat, tools, streaming]
+    status: active
+    tags: ["alibaba", "qwen", "cerebras", "fast"]
+    pricing:
+      input_per_token: 0.0000003
+      output_per_token: 0.0000003
+
+  # DeepSeek models
+  deepseek-r1-distill-llama-70b:
+    provider: cerebras
+    model_id: "deepseek-r1-distill-llama-70b"
+    display_name: "DeepSeek R1 Distill Llama 70B"
+    verification:
+      status: unverified
+      verified_at: "2026-01-29"
+      source: "documentation"
+    context_window: 8192
+    capabilities: [chat, tools, streaming, reasoning]
+    status: active
+    tags: ["deepseek", "distill", "cerebras", "fast", "reasoning"]
+    pricing:
+      input_per_token: 0.00000085
+      output_per_token: 0.00000085

package/v1/models/claude.yaml

@@ -0,0 +1,59 @@
+$schema: "https://raw.githubusercontent.com/ailib-official/ai-protocol/main/schemas/v1.json"
+protocol_version: "1.1"
+
+# Claude Model Instances
+models:
+  claude-3-5-sonnet:
+    provider: anthropic
+    model_id: "claude-3-5-sonnet-20241022"
+    display_name: "Claude 3.5 Sonnet"
+    verification:
+      status: unverified
+      source: "manual"
+    context_window: 200000
+    capabilities: [chat, vision, tools, streaming, agentic, reasoning]
+    pricing:
+      input_per_token: 0.000003
+      output_per_token: 0.000015
+    status: active
+    tags: ["claude", "vision", "reasoning", "agentic", "thinking"]
+    agentic_capabilities:
+      reasoning_effort: "high"
+      thinking_blocks: true
+      parallel_tools: false
+      builtin_tools: ["web_search", "document_analysis"]
+
+  claude-3-5-haiku:
+    provider: anthropic
+    model_id: "claude-3-5-haiku-20241022"
+    display_name: "Claude 3.5 Haiku"
+    verification:
+      status: unverified
+      source: "manual"
+    context_window: 200000
+    capabilities: [chat, vision, tools, streaming, agentic]
+    pricing:
+      input_per_token: 0.0000008
+      output_per_token: 0.000004
+    status: active
+    tags: ["claude", "vision", "fast", "cost-effective", "agentic"]
+
+  claude-3-opus:
+    provider: anthropic
+    model_id: "claude-3-opus-20240229"
+    display_name: "Claude 3 Opus"
+    verification:
+      status: unverified
+      source: "manual"
+    context_window: 200000
+    capabilities: [chat, vision, tools, streaming, agentic, reasoning]
+    pricing:
+      input_per_token: 0.000015
+      output_per_token: 0.000075
+    status: active
+    tags: ["claude", "vision", "premium", "reasoning", "agentic"]
+    agentic_capabilities:
+      reasoning_effort: "high"
+      thinking_blocks: true
+      parallel_tools: true
+      builtin_tools: ["web_search", "code_execution", "document_analysis"]

package/v1/models/cohere.yaml

@@ -0,0 +1,165 @@
+$schema: "https://raw.githubusercontent.com/ailib-official/ai-protocol/main/schemas/v1.json"
+protocol_version: "1.1"
+
+# Cohere Model Instances
+#
+# VERIFIED_AT: 2026-01-31 (based on public Cohere documentation)
+# FACT-CHECK 2026-02: https://docs.cohere.com/docs/models — command-r, command-r-plus
+#   deprecated Sept 15, 2025. Use command-r-plus-08-2024 or command-r-08-2024 (Live).
+#
+# Notes:
+# - Cohere is known for enterprise-grade LLMs.
+# - For Chat API use Live models: command-r-plus-08-2024, command-r-08-2024, command-a-03-2025, etc.
+
+models:
+  # Command R+ (deprecated Sept 15, 2025 — use command-r-plus-08-2024)
+  command-r-plus:
+    provider: cohere
+    model_id: "command-r-plus"
+    display_name: "Command R+ (deprecated)"
+    verification:
+      status: unverified
+      verified_at: "2026-02-04"
+      source: "https://docs.cohere.com/docs/models"
+      notes: "Deprecated Sept 15, 2025. Use command-r-plus-08-2024."
+    context_window: 131072
+    capabilities: [chat, tools, streaming]
+    status: deprecated
+    tags: ["cohere", "command", "flagship", "rag"]
+    pricing:
+      input_per_token: 0.000003
+      output_per_token: 0.000015
+
+  command-r-plus-08-2024:
+    provider: cohere
+    model_id: "command-r-plus-08-2024"
+    display_name: "Command R+ (08-2024)"
+    verification:
+      status: unverified
+      verified_at: "2026-01-31"
+      source: "documentation"
+    context_window: 131072
+    capabilities: [chat, tools, streaming]
+    status: active
+    tags: ["cohere", "command", "flagship"]
+    pricing:
+      input_per_token: 0.0000025
+      output_per_token: 0.00001
+
+  # Command R (deprecated Sept 15, 2025 — use command-r-08-2024)
+  command-r:
+    provider: cohere
+    model_id: "command-r"
+    display_name: "Command R (deprecated)"
+    verification:
+      status: unverified
+      verified_at: "2026-02-04"
+      source: "https://docs.cohere.com/docs/models"
+      notes: "Deprecated Sept 15, 2025. Use command-r-08-2024."
+    context_window: 131072
+    capabilities: [chat, tools, streaming]
+    status: deprecated
+    tags: ["cohere", "command", "standard"]
+    pricing:
+      input_per_token: 0.0000005
+      output_per_token: 0.0000015
+
+  command-r-08-2024:
+    provider: cohere
+    model_id: "command-r-08-2024"
+    display_name: "Command R (08-2024)"
+    verification:
+      status: unverified
+      verified_at: "2026-01-31"
+      source: "documentation"
+    context_window: 131072
+    capabilities: [chat, tools, streaming]
+    status: active
+    tags: ["cohere", "command"]
+    pricing:
+      input_per_token: 0.00000015
+      output_per_token: 0.0000006
+
+  # Command Light (legacy)
+  command-light:
+    provider: cohere
+    model_id: "command-light"
+    display_name: "Command Light"
+    verification:
+      status: unverified
+      verified_at: "2026-01-31"
+      source: "documentation"
+      notes: "Lightweight model, being deprecated in favor of Command R."
+    context_window: 4096
+    capabilities: [chat, streaming]
+    status: deprecated
+    tags: ["cohere", "command", "light", "legacy"]
+    pricing:
+      input_per_token: 0.0000003
+      output_per_token: 0.0000006
+
+  # Embed models
+  embed-english-v3.0:
+    provider: cohere
+    model_id: "embed-english-v3.0"
+    display_name: "Embed English v3"
+    verification:
+      status: unverified
+      verified_at: "2026-01-31"
+      source: "documentation"
+    context_window: 512
+    capabilities: [embedding]
+    status: active
+    tags: ["cohere", "embed", "english"]
+    pricing:
+      input_per_token: 0.0000001
+      output_per_token: 0
+
+  embed-multilingual-v3.0:
+    provider: cohere
+    model_id: "embed-multilingual-v3.0"
+    display_name: "Embed Multilingual v3"
+    verification:
+      status: unverified
+      verified_at: "2026-01-31"
+      source: "documentation"
+    context_window: 512
+    capabilities: [embedding]
+    status: active
+    tags: ["cohere", "embed", "multilingual"]
+    pricing:
+      input_per_token: 0.0000001
+      output_per_token: 0
+
+  # Rerank models
+  rerank-english-v3.0:
+    provider: cohere
+    model_id: "rerank-english-v3.0"
+    display_name: "Rerank English v3"
+    verification:
+      status: unverified
+      verified_at: "2026-01-31"
+      source: "documentation"
+    context_window: 4096
+    capabilities: [rerank]
+    status: active
+    tags: ["cohere", "rerank", "english"]
+    pricing:
+      input_per_token: 0.000002
+      output_per_token: 0
+
+  rerank-multilingual-v3.0:
+    provider: cohere
+    model_id: "rerank-multilingual-v3.0"
+    display_name: "Rerank Multilingual v3"
+    verification:
+      status: unverified
+      verified_at: "2026-01-31"
+      source: "documentation"
+    context_window: 4096
+    capabilities: [rerank]
+    status: active
+    tags: ["cohere", "rerank", "multilingual"]
+    pricing:
+      input_per_token: 0.000002
+      output_per_token: 0

package/v1/models/deepinfra.yaml

@@ -0,0 +1,29 @@
+$schema: ../../schemas/v1.json
+protocol_version: "1.1"
+models:
+  "deepinfra/meta-llama/Meta-Llama-3-8B-Instruct":
+    provider: "deepinfra"
+    model_id: "meta-llama/Meta-Llama-3-8B-Instruct"
+    verification:
+      status: unverified
+      source: "manual"
+    context_window: 8192
+    capabilities: ["chat"]
+
+  "deepinfra/mistralai/Mistral-7B-Instruct-v0.3":
+    provider: "deepinfra"
+    model_id: "mistralai/Mistral-7B-Instruct-v0.3"
+    verification:
+      status: unverified
+      source: "manual"
+    context_window: 32768
+    capabilities: ["chat", "tools"]
+    
+  "deepinfra/microsoft/WizardLM-2-8x22B":
+    provider: "deepinfra"
+    model_id: "microsoft/WizardLM-2-8x22B"
+    verification:
+      status: unverified
+      source: "manual"
+    context_window: 65536
+    capabilities: ["chat"]

package/v1/models/deepseek-chat.yaml

@@ -0,0 +1,32 @@
+$schema: "https://raw.githubusercontent.com/ailib-official/ai-protocol/main/schemas/v1.json"
+protocol_version: "1.1"
+
+models:
+  deepseek-chat:
+    provider: deepseek
+    model_id: "deepseek-chat"
+    display_name: "DeepSeek Chat"
+    verification:
+      status: verified
+      verified_at: "2026-01-28"
+      source: "list_models:/models"
+    context_window: 32768
+    capabilities: [chat, streaming, tools]
+    pricing:
+      input_per_token: 0.00000014 # example
+      output_per_token: 0.00000028 # example
+    status: active
+    tags: ["deepseek", "chat", "tools"]
+
+  deepseek-reasoner:
+    provider: deepseek
+    model_id: "deepseek-reasoner"
+    display_name: "DeepSeek Reasoner"
+    verification:
+      status: verified
+      verified_at: "2026-03-06"
+      source: "deepseek-console"
+    context_window: 64000
+    capabilities: [chat, streaming, tools, reasoning]
+    status: active
+    tags: ["deepseek", "reasoning", "tools"]

package/v1/models/doubao.yaml

@@ -0,0 +1,129 @@
+$schema: "https://raw.githubusercontent.com/ailib-official/ai-protocol/main/schemas/v1.json"
+protocol_version: "1.1"
+
+# 字节跳动豆包 (Doubao) Model Instances
+#
+# VERIFIED_AT: 2026-01-31 (based on public Volcengine documentation)
+#
+# Notes:
+# - Doubao (豆包) is ByteDance's flagship LLM series.
+# - Accessed via Volcengine (火山引擎) platform.
+# - Model IDs are endpoint IDs that need to be created in Volcengine console.
+
+models:
+  # Doubao Pro series
+  doubao-pro-32k:
+    provider: doubao
+    model_id: "doubao-pro-32k"
+    display_name: "豆包 Pro 32K"
+    verification:
+      status: unverified
+      verified_at: "2026-01-31"
+      source: "documentation"
+      notes: "Based on Volcengine documentation. Endpoint ID required."
+    context_window: 32768
+    capabilities: [chat, tools, streaming]
+    status: active
+    tags: ["bytedance", "doubao", "pro", "cn"]
+    pricing:
+      input_per_token: 0.0000008
+      output_per_token: 0.000002
+
+  doubao-pro-128k:
+    provider: doubao
+    model_id: "doubao-pro-128k"
+    display_name: "豆包 Pro 128K"
+    verification:
+      status: unverified
+      verified_at: "2026-01-31"
+      source: "documentation"
+    context_window: 131072
+    capabilities: [chat, tools, streaming]
+    status: active
+    tags: ["bytedance", "doubao", "pro", "long_context", "cn"]
+    pricing:
+      input_per_token: 0.000005
+      output_per_token: 0.000009
+
+  doubao-pro-256k:
+    provider: doubao
+    model_id: "doubao-pro-256k"
+    display_name: "豆包 Pro 256K"
+    verification:
+      status: unverified
+      verified_at: "2026-01-31"
+      source: "documentation"
+    context_window: 262144
+    capabilities: [chat, tools, streaming]
+    status: active
+    tags: ["bytedance", "doubao", "pro", "ultra_long_context", "cn"]
+    pricing:
+      input_per_token: 0.000005
+      output_per_token: 0.000009
+
+  # Doubao Lite series (cost-effective)
+  doubao-lite-32k:
+    provider: doubao
+    model_id: "doubao-lite-32k"
+    display_name: "豆包 Lite 32K"
+    verification:
+      status: unverified
+      verified_at: "2026-01-31"
+      source: "documentation"
+    context_window: 32768
+    capabilities: [chat, streaming]
+    status: active
+    tags: ["bytedance", "doubao", "lite", "cost_effective", "cn"]
+    pricing:
+      input_per_token: 0.0000003
+      output_per_token: 0.0000006
+
+  doubao-lite-128k:
+    provider: doubao
+    model_id: "doubao-lite-128k"
+    display_name: "豆包 Lite 128K"
+    verification:
+      status: unverified
+      verified_at: "2026-01-31"
+      source: "documentation"
+    context_window: 131072
+    capabilities: [chat, streaming]
+    status: active
+    tags: ["bytedance", "doubao", "lite", "long_context", "cn"]
+    pricing:
+      input_per_token: 0.0000008
+      output_per_token: 0.000001
+
+  # Doubao Vision
+  doubao-vision-pro-32k:
+    provider: doubao
+    model_id: "doubao-vision-pro-32k"
+    display_name: "豆包 Vision Pro 32K"
+    verification:
+      status: unverified
+      verified_at: "2026-01-31"
+      source: "documentation"
+    context_window: 32768
+    capabilities: [chat, vision, streaming]
+    status: active
+    tags: ["bytedance", "doubao", "vision", "multimodal", "cn"]
+    pricing:
+      input_per_token: 0.000003
+      output_per_token: 0.000009
+
+  # Doubao Embedding
+  doubao-embedding:
+    provider: doubao
+    model_id: "doubao-embedding"
+    display_name: "豆包 Embedding"
+    verification:
+      status: unverified
+      verified_at: "2026-01-31"
+      source: "documentation"
+    context_window: 4096
+    capabilities: [embedding]
+    status: active
+    tags: ["bytedance", "doubao", "embedding", "cn"]
+    pricing:
+      input_per_token: 0.0000005
+      output_per_token: 0