Model registrykdeps asset

kdeps-models-cloud

v1.0.0

kdeps model: cloud

set models / shelf model

PurposeThe model catalogs: downloadable llamafile and GGUF models, the cloud models of each provider with their limits, and the harvested OpenRouter free models.

LLM context: none - kdeps uses it without sending it to the model

Where it runs

Part of the models set, used in workflow and agent mode. The model catalogs: downloadable llamafile and GGUF models, the cloud models of each provider with their limits, and the harvested OpenRouter free models.

a chat resource or /model names a model
the registry resolves it: download, local server or providermodels/cloud
the model is called

Loaded by: kdeps (agent loop), /update, kdeps run, kdeps llamafile list, kdeps llm models, kdeps install, kdeps list, kdeps uninstall

In the code: kdeps (agent loop)llm.ReloadCloudModelsllm.readModelRegistry

Install

A kdeps asset: installed with the first-run profile that includes its set, or the first time a command needs it, and updated without a new release.

kdeps install models/cloud

Pin a version with kdeps install models/cloud@<version>.

49 cloud models

Model	Backend	Context	Output	Supports	Description
claude-opus-4-8	anthropic	1M	128k	thinking, images	most capable
claude-sonnet-4-6	anthropic	1M	64k	thinking, images	balanced speed/intelligence
claude-haiku-4-5-20251001	anthropic	200k	64k	thinking, images	fast and lightweight
gemini-2.5-pro	google	1M	8k	thinking, images	most capable, best reasoning
gemini-2.5-flash	google	1M	8k	thinking, images	fast multimodal
gemini-2.0-flash	google	1M	8k	images	balanced
gemini-1.5-pro	google	2.1M	8k	images	long context
gpt-4o	openai	128k	16k	images	flagship multimodal
gpt-4o-mini	openai	128k	16k	images	fast and cheap
o4-mini	openai	200k	16k	thinking, images	fast reasoning
o3	openai	200k	16k	thinking, images	advanced reasoning
o1	openai	200k	16k	thinking	reasoning
grok-3	xai	131k	4k		most capable
grok-3-fast	xai	131k	4k		fast
grok-3-mini	xai	131k	4k	thinking	small and cheap
grok-2	xai	131k	4k		previous generation
deepseek-v4-pro	deepseek	64k	8k		flagship
deepseek-v4-flash	deepseek	64k	8k		fast
llama-3.3-70b-versatile	groq	128k	8k		fast Llama 3.3 70B
llama-3.1-8b-instant	groq	128k	8k		fastest, smallest
gemma2-9b-it	groq	8k	4k		Google Gemma 2
mistral-large-latest	mistral	131k	8k		most capable
mistral-small-latest	mistral	33k	8k		fast and cheap
codestral-latest	mistral	256k	8k		code specialist
command-r-plus	cohere	128k	4k		most capable
command-r	cohere	128k	4k		balanced
meta-llama/Llama-3-70b-chat-hf	together	8k	4k		Llama 3 70B
meta-llama/Meta-Llama-3.1-8B-Instruct-Turbo	together	128k	4k		Llama 3.1 8B fast
llama-3.1-sonar-large-128k-online	perplexity	128k	4k		with web search
llama-3.1-sonar-small-128k-online	perplexity	128k	4k		fast with web search
m365-copilot	m365	128k	16k		default - service picks the model
quick	m365	128k	16k		fast, lower-latency responses
think-deeper	m365	128k	16k	thinking	extended reasoning
claude-sonnet	m365	200k	64k		Claude Sonnet via M365
claude-sonnet-think-deeper	m365	200k	64k	thinking	Claude Sonnet, extended reasoning
claude-opus	m365	200k	128k		Claude Opus via M365
gpt-5.6-think-deeper	m365	200k	16k	thinking	GPT-5.6, extended reasoning (reasoning only for now)
gpt-5.5	m365	200k	16k		GPT-5.5
gpt-5.5-quick	m365	200k	16k		GPT-5.5, fast
gpt-5.5-think-deeper	m365	200k	16k	thinking	GPT-5.5, extended reasoning
gpt-5.4	m365	200k	16k	thinking	GPT-5.4
gpt-5.4-quick	m365	200k	16k		GPT-5.4, fast
gpt-5.4-think-deeper	m365	200k	16k	thinking	GPT-5.4, extended reasoning
gpt-5.3	m365	128k	16k		GPT-5.3
gpt-5.3-quick	m365	128k	16k		GPT-5.3, fast
gpt-5.3-think-deeper	m365	128k	16k	thinking	GPT-5.3, extended reasoning
gpt-5.2	m365	128k	16k		GPT-5.2
gpt-5.2-quick	m365	128k	16k		GPT-5.2, fast
gpt-5.2-think-deeper	m365	128k	16k	thinking	GPT-5.2, extended reasoning

# Cloud model catalog: model IDs kdeps knows per provider, with their
# context window and output limits. Hand-maintained; bump version on change.
version: 1.0.0
models:
  # anthropic
  - id: "claude-opus-4-8"
    backend: anthropic
    desc: "most capable"
    envVar: ANTHROPIC_API_KEY
    thinking: true
    images: true
    contextWindow: 1000000
    maxOutputTokens: 128000
  - id: "claude-sonnet-4-6"
    backend: anthropic
    desc: "balanced speed/intelligence"
    envVar: ANTHROPIC_API_KEY
    thinking: true
    images: true
    contextWindow: 1000000
    maxOutputTokens: 64000
  - id: "claude-haiku-4-5-20251001"
    backend: anthropic
    desc: "fast and lightweight"
    envVar: ANTHROPIC_API_KEY
    thinking: true
    images: true
    contextWindow: 200000
    maxOutputTokens: 64000
  # google
  - id: "gemini-2.5-pro"
    backend: google
    desc: "most capable, best reasoning"
    envVar: GOOGLE_API_KEY
    thinking: true
    images: true
    contextWindow: 1048576
    maxOutputTokens: 8192
  - id: "gemini-2.5-flash"
    backend: google
    desc: "fast multimodal"
    envVar: GOOGLE_API_KEY
    thinking: true
    images: true
    contextWindow: 1048576
    maxOutputTokens: 8192
  - id: "gemini-2.0-flash"
    backend: google
    desc: "balanced"
    envVar: GOOGLE_API_KEY
    images: true
    contextWindow: 1048576
    maxOutputTokens: 8192
  - id: "gemini-1.5-pro"
    backend: google
    desc: "long context"
    envVar: GOOGLE_API_KEY
    images: true
    contextWindow: 2097152
    maxOutputTokens: 8192
  # openai
  - id: "gpt-4o"
    backend: openai
    desc: "flagship multimodal"
    envVar: OPENAI_API_KEY
    images: true
    contextWindow: 128000
    maxOutputTokens: 16384
  - id: "gpt-4o-mini"
    backend: openai
    desc: "fast and cheap"
    envVar: OPENAI_API_KEY
    images: true
    contextWindow: 128000
    maxOutputTokens: 16384
  - id: "o4-mini"
    backend: openai
    desc: "fast reasoning"
    envVar: OPENAI_API_KEY
    thinking: true
    images: true
    contextWindow: 200000
    maxOutputTokens: 16384
  - id: "o3"
    backend: openai
    desc: "advanced reasoning"
    envVar: OPENAI_API_KEY
    thinking: true
    images: true
    contextWindow: 200000
    maxOutputTokens: 16384
  - id: "o1"
    backend: openai
    desc: "reasoning"
    envVar: OPENAI_API_KEY
    thinking: true
    contextWindow: 200000
    maxOutputTokens: 16384
  # xai
  - id: "grok-3"
    backend: xai
    desc: "most capable"
    envVar: XAI_API_KEY
    contextWindow: 131072
    maxOutputTokens: 4096
  - id: "grok-3-fast"
    backend: xai
    desc: "fast"
    envVar: XAI_API_KEY
    contextWindow: 131072
    maxOutputTokens: 4096
  - id: "grok-3-mini"
    backend: xai
    desc: "small and cheap"
    envVar: XAI_API_KEY
    thinking: true
    contextWindow: 131072
    maxOutputTokens: 4096
  - id: "grok-2"
    backend: xai
    desc: "previous generation"
    envVar: XAI_API_KEY
    contextWindow: 131072
    maxOutputTokens: 4096
  # deepseek
  - id: "deepseek-v4-pro"
    backend: deepseek
    desc: "flagship"
    envVar: DEEPSEEK_API_KEY
    contextWindow: 64000
    maxOutputTokens: 8192
  - id: "deepseek-v4-flash"
    backend: deepseek
    desc: "fast"
    envVar: DEEPSEEK_API_KEY
    contextWindow: 64000
    maxOutputTokens: 8192
  # groq
  - id: "llama-3.3-70b-versatile"
    backend: groq
    desc: "fast Llama 3.3 70B"
    envVar: GROQ_API_KEY
    contextWindow: 128000
    maxOutputTokens: 8192
  - id: "llama-3.1-8b-instant"
    backend: groq
    desc: "fastest, smallest"
    envVar: GROQ_API_KEY
    contextWindow: 128000
    maxOutputTokens: 8192
  - id: "gemma2-9b-it"
    backend: groq
    desc: "Google Gemma 2"
    envVar: GROQ_API_KEY
    contextWindow: 8192
    maxOutputTokens: 4096
  # mistral
  - id: "mistral-large-latest"
    backend: mistral
    desc: "most capable"
    envVar: MISTRAL_API_KEY
    contextWindow: 131072
    maxOutputTokens: 8192
  - id: "mistral-small-latest"
    backend: mistral
    desc: "fast and cheap"
    envVar: MISTRAL_API_KEY
    contextWindow: 32768
    maxOutputTokens: 8192
  - id: "codestral-latest"
    backend: mistral
    desc: "code specialist"
    envVar: MISTRAL_API_KEY
    contextWindow: 256000
    maxOutputTokens: 8192
  # cohere
  - id: "command-r-plus"
    backend: cohere
    desc: "most capable"
    envVar: COHERE_API_KEY
    contextWindow: 128000
    maxOutputTokens: 4096
  - id: "command-r"
    backend: cohere
    desc: "balanced"
    envVar: COHERE_API_KEY
    contextWindow: 128000
    maxOutputTokens: 4096
  # together
  - id: "meta-llama/Llama-3-70b-chat-hf"
    backend: together
    desc: "Llama 3 70B"
    envVar: TOGETHER_API_KEY
    contextWindow: 8192
    maxOutputTokens: 4096
  - id: "meta-llama/Meta-Llama-3.1-8B-Instruct-Turbo"
    backend: together
    desc: "Llama 3.1 8B fast"
    envVar: TOGETHER_API_KEY
    contextWindow: 128000
    maxOutputTokens: 4096
  # perplexity
  - id: "llama-3.1-sonar-large-128k-online"
    backend: perplexity
    desc: "with web search"
    envVar: PERPLEXITY_API_KEY
    contextWindow: 128000
    maxOutputTokens: 4096
  - id: "llama-3.1-sonar-small-128k-online"
    backend: perplexity
    desc: "fast with web search"
    envVar: PERPLEXITY_API_KEY
    contextWindow: 128000
    maxOutputTokens: 4096
  # m365
  - id: "m365-copilot"
    backend: m365
    desc: "default - service picks the model"
    contextWindow: 128000
    maxOutputTokens: 16384
  - id: "quick"
    backend: m365
    desc: "fast, lower-latency responses"
    contextWindow: 128000
    maxOutputTokens: 16384
  - id: "think-deeper"
    backend: m365
    desc: "extended reasoning"
    thinking: true
    contextWindow: 128000
    maxOutputTokens: 16384
  - id: "claude-sonnet"
    backend: m365
    desc: "Claude Sonnet via M365"
    contextWindow: 200000
    maxOutputTokens: 64000
  - id: "claude-sonnet-think-deeper"
    backend: m365
    desc: "Claude Sonnet, extended reasoning"
    thinking: true
    contextWindow: 200000
    maxOutputTokens: 64000
  - id: "claude-opus"
    backend: m365
    desc: "Claude Opus via M365"
    contextWindow: 200000
    maxOutputTokens: 128000
  - id: "gpt-5.6-think-deeper"
    backend: m365
    desc: "GPT-5.6, extended reasoning (reasoning only for now)"
    thinking: true
    contextWindow: 200000
    maxOutputTokens: 16384
  - id: "gpt-5.5"
    backend: m365
    desc: "GPT-5.5"
    contextWindow: 200000
    maxOutputTokens: 16384
  - id: "gpt-5.5-quick"
    backend: m365
    desc: "GPT-5.5, fast"
    contextWindow: 200000
    maxOutputTokens: 16384
  - id: "gpt-5.5-think-deeper"
    backend: m365
    desc: "GPT-5.5, extended reasoning"
    thinking: true
    contextWindow: 200000
    maxOutputTokens: 16384
  - id: "gpt-5.4"
    backend: m365
    desc: "GPT-5.4"
    thinking: true
    contextWindow: 200000
    maxOutputTokens: 16384
  - id: "gpt-5.4-quick"
    backend: m365
    desc: "GPT-5.4, fast"
    contextWindow: 200000
    maxOutputTokens: 16384
  - id: "gpt-5.4-think-deeper"
    backend: m365
    desc: "GPT-5.4, extended reasoning"
    thinking: true
    contextWindow: 200000
    maxOutputTokens: 16384
  - id: "gpt-5.3"
    backend: m365
    desc: "GPT-5.3"
    contextWindow: 128000
    maxOutputTokens: 16384
  - id: "gpt-5.3-quick"
    backend: m365
    desc: "GPT-5.3, fast"
    contextWindow: 128000
    maxOutputTokens: 16384
  - id: "gpt-5.3-think-deeper"
    backend: m365
    desc: "GPT-5.3, extended reasoning"
    thinking: true
    contextWindow: 128000
    maxOutputTokens: 16384
  - id: "gpt-5.2"
    backend: m365
    desc: "GPT-5.2"
    contextWindow: 128000
    maxOutputTokens: 16384
  - id: "gpt-5.2-quick"
    backend: m365
    desc: "GPT-5.2, fast"
    contextWindow: 128000
    maxOutputTokens: 16384
  - id: "gpt-5.2-think-deeper"
    backend: m365
    desc: "GPT-5.2, extended reasoning"
    thinking: true
    contextWindow: 128000
    maxOutputTokens: 16384

Versions

Version	Published	Status
1.0.0	10/13/2026	active

Details

Author: kdeps
License: Apache-2.0
Latest Version: 1.0.0
Published: 10/13/2026

kdeps-models-cloud

Where it runs

Install

Contents

Versions

Details

Tags