Skip to main content

​
AI Chat Models

MachinaOs ships 11 chat-model nodes, with models fetched dynamically from each provider’s API (and, for the local nodes, from your running server). The backend uses a hybrid architecture: a native SDK layer in server/services/llm/ for direct chat completions, and a LangChain + LangGraph path for agent tool-calling.

​
Available Chat-Model Nodes

NodeModelsBest For
OpenAIGPT-5.5/5.4/5.2, GPT-4.1/4.1-mini/nano, o3, o4-mini, GPT-4o/4o-miniGeneral purpose, reasoning
AnthropicClaude Fable 5, Opus 4.8/4.7, Sonnet 4.6, Haiku 4.5Coding, analysis, extended thinking
Google GeminiGemini 3.5-flash, 3.1-pro/flash-lite, 3-flash, 2.5-pro/flash/flash-liteMultimodal, long context (1M)
OpenRouter200+ modelsAccess multiple providers via single API
GroqLlama 3.3-70b, Llama 3.1-8b, GPT-OSS-120b/20b, Qwen3-32bUltra-fast inference (LPU)
CerebrasGPT-OSS-120b, Qwen-3-235b, GLM-4.7Ultra-fast on wafer-scale hardware
DeepSeekdeepseek-v4-flash, deepseek-v4-proV4 with reasoning modes, up to 1M context
Kimikimi-k2.6, kimi-k2.5, kimi-k2.7-code256K context, thinking on by default
Mistralmistral-large/medium/small-latest, codestral-latestUp to 256K context
OllamaWhatever you’ve pulled locallyLocal models, no API key
LM StudioWhatever you’ve loaded locallyLocal models, no API key
xAI (Grok) is not a chat-model node. It is available only through the internal native-chat path (the shared OpenAI-compatible provider with base_url=https://api.x.ai/v1). The two nodes new since older docs are Ollama and LM Studio, which connect to a local server running on your machine.

​
Adding API Keys

  1. Click the key icon in the toolbar
  2. Select the provider
  3. Enter your API key
  4. Click Validate to test
API keys are encrypted and stored locally. They’re never sent to MachinaOs servers.

​
OpenAI Chat Model

​
Models

ModelBest For
gpt-5.5Most capable, up to ~1M context
gpt-5.4High capability, long context
gpt-5.2Capable, 400K context
gpt-4.1 / 4.1-mini / 4.1-nanoFast, 1M context
o3Advanced reasoning
o4-miniFast, efficient reasoning
gpt-4o / gpt-4o-miniMultimodal, cost-effective

​
Parameters

​
model
select
required
The model to use
​
prompt
string
required
The message to send. Supports template variables.
​
temperature
slider
default:"0.7"
Randomness (0 = deterministic, 1 = creative)
​
maxTokens
number
default:"1000"
Maximum response length
​
responseFormat
select
default:"text"
Output format: text or json_object
​
reasoningEffort
select
default:"medium"
For o-series and GPT-5 hybrid reasoning: low, medium, high (GPT-5 also supports xhigh)

​
Output

{
  "response": "The AI's response text",
  "model": "gpt-5.5",
  "thinking": "Reasoning process (o-series / GPT-5 only)",
  "usage": {
    "prompt_tokens": 50,
    "completion_tokens": 100,
    "total_tokens": 150
  }
}

​
Anthropic Claude Model

​
Models

ModelBest For
claude-opus-4-8Most capable, detailed analysis, 1M context
claude-opus-4-7High capability, 1M context
claude-sonnet-4-6Best for coding and complex tasks, 1M context
claude-fable-5Capable, 1M context
claude-haiku-4-5Fast responses, simple tasks, 200K context

​
Parameters

​
model
select
required
Claude model to use
​
prompt
string
required
The message to send
​
systemPrompt
string
System instructions for the model
​
temperature
slider
default:"0.7"
Randomness (0-1)
​
maxTokens
number
default:"1000"
Maximum response length
​
thinkingEnabled
boolean
default:"false"
Enable extended thinking mode (all Claude 4.x models)
​
thinkingBudget
number
default:"2048"
Token budget for thinking (1024-16000). Shown when thinkingEnabled is true.

​
Extended Thinking

Claude’s extended thinking mode shows the model’s reasoning process: 
{
  "response": "Claude's final response",
  "thinking": "Let me analyze this step by step...",
  "model": "claude-opus-4-8",
  "stop_reason": "end_turn"
}
Anthropic model IDs use hyphens, not dots (claude-sonnet-4-6, not claude-sonnet-4.6).
When thinking is enabled, max_tokens must be greater than thinkingBudget. Temperature is automatically set to 1.

​
Google Gemini Model

​
Models

ModelBest For
gemini-3.5-flashFast, frontier performance, 1M context
gemini-3.1-pro-previewMost intelligent, complex tasks
gemini-3.1-flash-liteFast, cost-effective
gemini-3-flash-previewFast, multimodal
gemini-2.5-proComplex tasks, thinking support
gemini-2.5-flash / 2.5-flash-liteFast, thinking support

​
Parameters

​
model
select
required
Gemini model to use
​
prompt
string
required
The message to send
​
temperature
slider
default:"0.7"
Randomness (0-1)
​
maxTokens
number
default:"1000"
Maximum response length
​
safetySettings
select
default:"default"
Content safety level
​
thinkingEnabled
boolean
default:"false"
Enable thinking mode (Gemini 3.x and 2.5 models)

​
Output

{
  "response": "Gemini's response",
  "thinking": "Reasoning process (when enabled)",
  "model": "gemini-3.5-flash"
}

​
OpenRouter Model

OpenRouter provides access to 200+ models from multiple providers through a single API.

​
Features

  • Unified API: One API key for OpenAI, Anthropic, Google, Meta, Mistral, and more
  • Free Models: Some models available at no cost (marked with [FREE] prefix)
  • Fallback: Automatic model fallback if primary is unavailable

​
Models

Models are grouped by cost in the dropdown:
  • Free models: [FREE] prefix, no cost
  • Paid models: Standard pricing per provider
Popular models include:
  • openai/gpt-5.5
  • anthropic/claude-sonnet-4.6 (the OpenRouter default)
  • google/gemini-3.5-flash
  • meta-llama/llama-3.3-70b-instruct
  • mistralai/mistral-large-latest

​
Parameters

​
model
select
required
Model in format: provider/model-name
​
prompt
string
required
The message to send
​
temperature
slider
default:"0.7"
Randomness (0-1)
​
maxTokens
number
default:"1000"
Maximum response length

​
Output

{
  "response": "Model's response",
  "model": "openai/gpt-5.5",
  "provider": "openrouter"
}

​
Groq Model

Groq provides ultra-fast inference on custom LPU (Language Processing Unit) hardware.

​
Models

ModelBest For
llama-3.3-70b-versatileGeneral purpose, fast
llama-3.1-8b-instantUltra-fast, simple tasks
openai/gpt-oss-120bLarge open model, long output
openai/gpt-oss-20bFast open model
qwen/qwen3-32bReasoning with parsed/hidden output

​
Parameters

​
model
select
required
Groq model to use
​
prompt
string
required
The message to send
​
temperature
slider
default:"0.7"
Randomness (0-1)
​
maxTokens
number
default:"1000"
Maximum response length
​
reasoningFormat
select
default:"parsed"
For Qwen3 models: “parsed” returns reasoning, “hidden” returns only final answer

​
Reasoning Output

Qwen3-32b supports reasoning output: 
{
  "response": "The final answer",
  "thinking": "Step-by-step reasoning process",
  "model": "qwen/qwen3-32b"
}

​
Cerebras Model

Cerebras provides ultra-fast inference on custom wafer-scale AI hardware.

​
Models

ModelBest For
gpt-oss-120bCapable, balanced (the default)
qwen-3-235bReasoning tasks (budget thinking)
zai-glm-4.7General purpose

​
Parameters

​
model
select
required
Cerebras model to use
​
prompt
string
required
The message to send
​
temperature
slider
default:"0.7"
Randomness (0-1)
​
maxTokens
number
default:"1000"
Maximum response length

​
Output

{
  "response": "Cerebras model response",
  "model": "gpt-oss-120b"
}

​
DeepSeek

DeepSeek V4 models with reasoning support and long context.

​
Models

ModelBest For
deepseek-v4-flashGeneral purpose, fast, up to 1M context, up to 64K output
deepseek-v4-proHigher capability, up to 1M context, up to 64K output
The legacy deepseek-chat and deepseek-reasoner aliases still resolve (scheduled for deprecation 2026-07-24). deepseek-reasoner returns always-on reasoning in the thinking field.

​
Parameters

​
model
select
required
DeepSeek model
​
prompt
string
required
The message to send
​
temperature
slider
default:"0.7"
Randomness (0-2)
​
maxTokens
number
default:"8192"
Maximum response length (up to 64K)

​
Reasoning Output

DeepSeek’s reasoning is mapped to the standard thinking field: 
{
  "response": "The final answer",
  "thinking": "Step-by-step reasoning",
  "model": "deepseek-v4-pro"
}

​
Kimi (Moonshot AI)

Moonshot’s Kimi models with 256K context window and thinking on by default.

​
Models

ModelBest For
kimi-k2.6General purpose, 256K context, 32K output
kimi-k2.5General purpose, 256K context, 96K output
kimi-k2.7-codeCode tasks, 256K context, 96K output

​
Parameters

​
model
select
required
Kimi model
​
prompt
string
required
The message to send
​
maxTokens
number
default:"4096"
Maximum response length (up to 96K)
Kimi models use a fixed temperature of 0.6. User-set temperature is ignored. Thinking is on by default but explicitly disabled in tool-calling agent mode for compatibility.

​
Output

{
  "response": "Kimi response",
  "thinking": "Reasoning (when thinking is active)",
  "model": "kimi-k2.6"
}

​
Mistral

Mistral AI models including Large, Small, and Codestral for code tasks.

​
Models

ModelBest For
mistral-large-latestMost capable, general purpose, 256K context
mistral-medium-latestBalanced, 256K context
mistral-small-latestFast, cost-effective
codestral-latestCode generation and completion, 256K context

​
Parameters

​
model
select
required
Mistral model
​
prompt
string
required
The message to send
​
temperature
slider
default:"0.7"
Randomness (0-1.5)
​
maxTokens
number
default:"8192"
Maximum response length (up to 131K)

​
Output

{
  "response": "Mistral response",
  "model": "mistral-large-latest"
}
Mistral models support up to 256K context but do not have a thinking mode. Temperature range is 0-1.5 (not 0-2).

​
Ollama (Local)

Run models locally through an Ollama server. No API key is needed; instead you point MachinaOs at your running server’s URL in the Credentials Modal (default http://localhost:11434/v1).

​
Models

The dropdown reflects whatever you have pulled into Ollama (e.g. qwen2.5, llama3.x, phi-3, deepseek-r1). MachinaOs probes your running server via the official Ollama Python SDK and reads per-model context length directly. When the server is offline the list is empty, cueing you to start it.

​
Parameters

​
model
select
required
A model loaded in your local Ollama server
​
prompt
string
required
The message to send
​
temperature
slider
default:"0.7"
Randomness (0-2)
​
maxTokens
number
default:"4096"
Maximum response length (per-model, roughly context / 4, capped at 4096)
Traffic stays on your machine. MachinaOs sends requests to your local server URL, never to a cloud API.

​
LM Studio (Local)

Run models locally through the LM Studio server. Like Ollama, no API key is required; you set your local server URL in the Credentials Modal (default http://localhost:1234/v1).

​
Models

The dropdown reflects whatever you have loaded in the LM Studio UI. MachinaOs probes the running server via the official LM Studio Python SDK and reads typed model info (context length, tool-use support, vision).

​
Parameters

​
model
select
required
A model loaded in your LM Studio server
​
prompt
string
required
The message to send
​
temperature
slider
default:"0.7"
Randomness (0-2)
​
maxTokens
number
default:"4096"
Maximum response length (per-model, roughly context / 4, capped at 4096)

​
Native SDK vs LangChain Path

MachinaOs uses a hybrid architecture for LLM access:
  • Native SDK path (server/services/llm/): Used by execute_chat() for direct chat completions. Returns a normalized LLMResponse across providers. This path also serves the OpenAI-compatible providers including DeepSeek, Kimi, Mistral, xAI, Ollama, and LM Studio.
  • LangChain path: Used by execute_agent() and execute_chat_agent() for tool-calling agents via LangGraph. All chat-model providers are supported (Groq and Cerebras use this path for direct chat too).
OpenAI-compatible providers (DeepSeek, Kimi, Mistral, xAI, Ollama, LM Studio) reuse the OpenAIProvider class with base_url read from server/config/llm_defaults.json. For the local nodes (Ollama, LM Studio) the base_url resolves to your machine’s server URL, so traffic never leaves your host. Adding a new OpenAI-compatible provider is largely a config change.

​
Thinking/Reasoning Modes

Several providers support extended thinking or reasoning modes that show the model’s internal reasoning process.
ProviderModelsParameter
ClaudeClaude 4.x modelsthinkingBudget (tokens)
GeminiGemini 3.x, 2.5 Pro/FlashthinkingBudget (tokens)
OpenAIo3, o4-mini, GPT-5.x hybridreasoningEffort (low/medium/high; GPT-5 also xhigh)
GroqQwen3-32breasoningFormat (parsed/hidden)
CerebrasQwen-3-235bbudget thinking
DeepSeekdeepseek-v4 (legacy reasoner alias)reasoning modes
Kimikimi-k2.6 / k2.5 / k2.7-codeOn by default (off in agent mode)

​
Using Thinking Output

The thinking field is available in the node output for downstream nodes: 
{{openaiChatModel.thinking}}
{{anthropicChatModel.thinking}}

​
Comparing Providers

FeatureOpenAIClaudeGeminiOpenRouterGroqCerebrasDeepSeekKimiMistral
SpeedFastMediumFastVariesUltra-fastUltra-fastFastFastFast
Reasoningo-series, GPT-5Extended thinkingThinking modeModel-dependentQwen3Qwen-3v4 reasoningOn by default-
Context Window128K-1M200K-1M1MVaries131K131Kup to 1M256Kup to 256K
MultimodalYesYesYesModel-dependentNoNoNoNoNo
JSON ModeYesNoNoModel-dependentNoNoYesYesYes
Ollama and LM Studio are local nodes: capabilities depend entirely on the model you have pulled/loaded.

​
Common Use Cases

​
Text Generation

Prompt: Write a product description for: {{input.product_name}}
Temperature: 0.8

​
Data Extraction

Prompt: Extract the email and phone from: {{input.text}}
Response Format: json_object
Temperature: 0

​
Complex Reasoning (with thinking)

Model: claude-sonnet-4-6
Thinking Enabled: true
Thinking Budget: 4096
Prompt: Analyze this code and explain the bug: {{input.code}}

​
Tips

Use temperature 0 for deterministic outputs like data extraction.
Use temperature 0.7-0.9 for creative writing tasks.
Enable thinking mode for complex reasoning tasks that benefit from step-by-step analysis.
Use OpenRouter to experiment with different models without managing multiple API keys.
API calls cost money. Monitor your usage in your provider’s dashboard.

​
Error Handling

ErrorCauseSolution
401 UnauthorizedInvalid API keyCheck/update API key
429 Rate LimitedToo many requestsAdd delay, reduce frequency
500 Server ErrorProvider issueRetry later

AI Agent

Use models with memory and tools

AI Skills

Extend Chat Agent capabilities

AI Tools

Tool nodes for AI agents

AI Tutorial

Build an AI-powered workflow