Home / Araçlar

Platform Tools

Built-in Tools

Extend model capabilities with powerful built-in tools. Give your models the ability to search the web, execute code, read files, control computers, and connect to external services — all within a single API call.

Responses API Required — Built-in tools are available through the Responses API (client.responses.create()). The Chat Completions API supports function calling only.

All Built-in Tools

Web Search

GA

Give models access to real-time web search results. Models can search the web, read pages, and synthesize information with citations automatically.

Real-time data Citations Bing-powered
Learn more →

File Search

GA

Semantic search over your own documents and data using vector stores. Supports up to 10,000 files per vector store with automatic chunking and embedding.

Vector stores RAG 50+ formats
Learn more →

Code Interpreter

GA

Execute Python code in a sandboxed environment. Analyze data, generate charts, perform mathematical computations, and process files — all securely isolated.

Python sandbox Data analysis Charts
Learn more →

Computer Use

Beta

Enable models to interact with computer interfaces — browsers, desktops, and web apps. Take screenshots, click, type, and navigate GUIs autonomously.

Browser control GUI automation Screenshots
Learn more →

MCP & Connectors

Beta

Connect to remote MCP servers and GlomaxGPT-maintained connectors for Slack, GitHub, Gmail, Salesforce, and more. Extend tool capabilities infinitely.

MCP protocol Slack GitHub Gmail
Learn more →

Image Generation

GA

Generate, edit, and transform images directly within a conversation. Powered by GlomaxGPT Vision 2 with support for masks, inpainting, and style control.

GlomaxGPT Vision 2 GlomaxGPT Vision 2 Inpainting
Learn more →

Apply Patch

Preview

Apply unified diffs and patches to code files. Ideal for agentic coding workflows where the model generates and applies code changes incrementally.

Unified diff Agentic coding
Learn more →

Local Shell

Preview

Execute shell commands on the host machine in agentic workflows. Run scripts, manage files, install packages, and interact with the OS directly.

Shell commands File management
Learn more →

Skills

Experimental

Reusable tool bundles that agents can discover and invoke. Define complex multi-step capabilities as named skills and share them across agents.

Reusable Multi-step
Learn more →

Tool Search

Experimental

Automatically discover and select the right tool from a large tool registry. Models search through hundreds of available tools to find the best match for a task.

Auto-discovery Tool registry
Learn more →

Tool Selection Guide

Use this table to quickly identify which tool fits your use case.

Use Case Recommended Tool Notes
Answer questions about recent events Web Search Use search_context_size: "high" for complex queries
Search company documents / knowledge base File Search Upload to a vector store first; supports 50+ file formats
Analyze CSV, run calculations Code Interpreter Runs in isolated Python sandbox; no internet access
Automate a web app or GUI Computer Use Beta; review security considerations before deploying
Connect to Slack / GitHub / Gmail MCP Connectors OAuth 2.0 authentication required per service
Generate or edit images Image Generation Powered by GlomaxGPT Vision 2; supports masks and inpainting
Apply code changes in agentic workflow Apply Patch Works with unified diff format
Run OS commands in agent Local Shell Preview; use with strict sandboxing

Adding Tools with Python

Tools are passed as a list in the tools parameter of the Responses API. You can combine multiple tools in a single request.

Python
from GlomaxGPT import GlomaxGPT



client = GlomaxGPT()



response = client.responses.create(



    model="glomaxgpt-ultra",



    input="What are the latest AI news today?",



    tools=[{"type": "web_search_preview"}]



)



print(response.output_text)
Python
from GlomaxGPT import GlomaxGPT



client = GlomaxGPT()



response = client.responses.create(



    model="glomaxgpt-ultra",



    input="Search recent papers and run the analysis code",



    tools=[



        {"type": "web_search_preview"},



        {"type": "code_interpreter", "container": {"type": "auto"}},



        {"type": "file_search", "vector_store_ids": ["vs_abc123"]}



    ]



)



print(response.output_text)
Python
from GlomaxGPT import GlomaxGPT



client = GlomaxGPT()



# Define a custom function tool



get_weather_tool = {



    "type": "function",



    "name": "get_weather",



    "description": "Returns current weather for a city",



    "parameters": {



        "type": "object",



        "properties": {



            "city": {"type": "string", "description": "City name"}



        },



        "required": ["city"]



    }



}



response = client.responses.create(



    model="glomaxgpt-ultra",



    input="What's the weather in Istanbul?",



    tools=[get_weather_tool]



)



print(response.output)
Pricing Note — Each tool invocation may incur additional costs beyond token usage. Web Search charges per search call, Code Interpreter charges per session, and File Search charges based on vector store storage. See the Pricing page for details.

Ready to build with tools?

Start with the Quickstart guide or explore individual tool documentation to integrate AI capabilities into your application.

BaÅŸla Playground'da Dene