Docs Check here before you buy

Tools overview

Which tools work with DaLAi, in five categories, and exactly where to configure each one.

DaLAi is compatible with the standard OpenAI API and also answers the Anthropic Messages format. Any tool that understands one of these two and lets you enter the base URL and key yourself works with DaLAi.

In most tools you change only two things: the base URL, to https://dalai.ir/v1, and the key, to a key you created in the Console. Tools that work with the Anthropic format take the URL without /v1; this exception is noted next to the tool in question.

The tools are grouped into five categories:

Coding tools: assistants and agents that write code alongside you in the editor or on the command line.
Agentic tools: ready-made agents, and the frameworks and libraries you use to build your own agent or product.
Automation and workflows: these put the model into everyday processes, mostly without coding.
Enterprise tools: multi-user platforms that an organization installs on its own server.
Personal chat apps: desktop and mobile apps for one person's daily work.

DaLAi does not support the Responses format. If a tool lets you choose between several formats, pick OpenAI Compatible or Chat Completions, not an option named Responses.

Coding tools

Agent means the tool itself reads files, changes them and runs commands, and you review the result. Assistant means it makes suggestions and every step is your call.

Tool	Where to configure
Claude Code Agent	Three environment variables, not the app's settings; URL without `/v1`. Details below
opencode Agent	An OpenAI-compatible provider in the `opencode.json` file. Full details in Coding with DaLAi
Cline Agent	Set API Provider to OpenAI Compatible, then Base URL and key
Kilo Code Agent	Settings → Providers → Custom provider; in Provider API choose OpenAI Compatible, and in Base URL enter `https://dalai.ir/v1`
Aider Assistant	`aider --openai-api-base https://dalai.ir/v1 --openai-api-key sk-...`
Continue Assistant	In `config.yaml`: `provider: openai` and `apiBase: https://dalai.ir/v1`
Zed Assistant and agent	Settings → AI → LLM Providers → Add Provider, type OpenAI, and API URL set to `https://dalai.ir/v1`

Fuller details, including choosing a model for code and keeping costs under control, are in Coding with DaLAi.

Claude Code

Claude Code works with the Anthropic format, and its setup differs from the rest of this page: it takes three environment variables, and unlike the others, you must not add /v1 to the URL. Ready-made commands for Windows, macOS and Linux are in Coding with DaLAi.

The first message of every Claude Code session costs much more than an ordinary message, because the tool sends a large system prompt before your work. Do not start with a balance close to zero. The measured figure is given on that page.

Agentic tools

An agent carries out a multi-step task on its own: it plans, calls tools, reads the result and takes the next step. DaLAi provides the model, and tool calling is supported in both formats. The tools themselves and MCP servers run on your device or server. Coding agents are in the previous category.

Ready-made agents

Tool	Use and where to configure
Goose	General-purpose agent for desktop and the command line, with MCP extensions. Settings → Models → Configure providers → Add Provider; set Provider Type to OpenAI Compatible and API URL to `https://dalai.ir/v1`

Frameworks for building agents

Claude Agent SDK, Python and TypeScript: the same variables as Claude Code. ANTHROPIC_BASE_URL set to https://dalai.ir without /v1, the key in ANTHROPIC_AUTH_TOKEN, and always specify the model name.
OpenAI Agents SDK, Python: OpenAIChatCompletionsModel with a client whose base_url is https://dalai.ir/v1. Example below.
LangChain and LangGraph, Python: ChatOpenAI with base_url set to https://dalai.ir/v1 and use_responses_api=False.
CrewAI, Python: LLM with model="openai/claude-sonnet-4-5" and base_url set to https://dalai.ir/v1.
Pydantic AI, Python: OpenAIChatModel with an OpenAIProvider whose base_url is https://dalai.ir/v1.

An example with OpenAI Agents SDK, which calls the Responses format without this setting:

from openai import AsyncOpenAI
from agents import Agent, Runner, OpenAIChatCompletionsModel, set_tracing_disabled

set_tracing_disabled(True)
client = AsyncOpenAI(base_url="https://dalai.ir/v1", api_key="sk-YOUR-KEY")
model = OpenAIChatCompletionsModel(model="claude-sonnet-4-5", openai_client=client)

agent = Agent(name="assistant", instructions="Answer briefly.", model=model)
print(Runner.run_sync(agent, "Hello").final_output)

The set_tracing_disabled(True) line turns off trace reporting so that the library does not try to send it to OpenAI's servers.

Programming libraries

OpenAI SDK, Python and Node: base_url or baseURL set to https://dalai.ir/v1. Example in API overview.
Anthropic SDK, Python and TypeScript: base_url set to https://dalai.ir, without /v1. For the claude-* models.
Vercel AI SDK, TypeScript: createOpenAICompatible from the @ai-sdk/openai-compatible package, with baseURL set to https://dalai.ir/v1.
LlamaIndex, Python: OpenAILike from the llama-index-llms-openai-like package, with api_base set to https://dalai.ir/v1 and is_chat_model=True.

If you get an error with one of these libraries, check whether the request goes to /v1/chat/completions. Some libraries call a different path by default: in LangChain set use_responses_api=False, in Pydantic AI use OpenAIChatModel, in Vercel AI SDK use the @ai-sdk/openai-compatible package, and in LlamaIndex set is_chat_model=True.

Automation and workflows

These tools put the model into a process: an event such as a new email, a form or a message calls the model, and the result goes on to the next system. You can install all four tools on your own server.

Tool	Use and where to configure
n8n	Workflow automation, with an AI Agent node. An OpenAI credential with Base URL set to `https://dalai.ir/v1`, then an OpenAI Chat Model node with Use Responses API turned off
Dify	Build apps, agents and workflows with a visual interface. Integrations → Model Provider, the OpenAI-API-compatible plugin with API Base URL set to `https://dalai.ir/v1`
Flowise	Build chains and agents with drag and drop. The OpenAI Custom Model node; enter the model name, and under Additional Parameters set Base Path to `https://dalai.ir/v1`
Langflow	Build flows and agents visually. Settings → Model Providers → OpenAI Compatible, with Base URL and key

Two settings that, if missed, stop the agent from working. In n8n, the Use Responses API option is on by default in a new node and you must turn it off; also, do not use the separate OpenAI node with the Message a Model action. In Dify, for each model set Function Call Type to Tool Call, and set Model context size to the model's actual context size.

Enterprise tools

Multi-user platforms that an organization installs on its own server. The DaLAi key stays on the server only, and staff sign in with their own organizational accounts. Which one to choose for which job, and how to set it up, is covered in the Enterprise usage guide.

Tool	Use and where to configure
Open WebUI	Internal chat portal with single sign-on (OIDC and LDAP), user groups and model access per group. Administrator: Settings → Admin → Connections, an OpenAI connection with `https://dalai.ir/v1`
LibreChat	Multi-user chat portal with single sign-on and a credit cap for each user. A custom endpoint in `librechat.yaml` with `baseURL` set to `https://dalai.ir/v1`
AnythingLLM	Chat with the organization's documents in separate workspaces, with user roles in the Docker version. Settings → AI Providers → LLM, the Generic OpenAI provider with Base URL set to `https://dalai.ir/v1`
LiteLLM Proxy	The organization's internal gateway: a separate key and budget for each team, behind a single DaLAi key. In `config.yaml`, with the `openai/` prefix on the model name and `api_base` set to `https://dalai.ir/v1`; example in the Enterprise usage guide

In Open WebUI, the models of a new connection are visible only to the administrator until the administrator makes them public or assigns them to a group. Chatting with documents requires an embedding model, which DaLAi does not offer yet; use the platforms' own local embedding model.

Personal chat apps

Tool	Use and where to configure
Cherry Studio	Desktop; chat, ready-made assistants and agents. Settings → Model Provider → Add Provider, full details below
Jan	Desktop; simple, and on your own device. Settings → Model Providers → Add Provider, the OpenAI-compatible format and Base URL set to `https://dalai.ir/v1`
Chatbox	Desktop and mobile. Settings → Model Provider → Add, OpenAI API Compatible mode and API Host set to `https://dalai.ir/v1`

In Jan, create a new provider and do not change the URL of the built-in OpenAI provider; that one calls the Responses format.

Cherry Studio

According to its makers, Cherry Studio is an "AI productivity studio with smart chat, autonomous agents, and 300+ assistants", with "unified access to frontier LLMs". It runs on Windows, macOS and Linux, and the app is installed on your own device. These steps were written for version 2.1.4.

Download the app from the Cherry Studio website and run it. If a provider selection screen opens on first launch, click Set up later.
Go to Settings with the gear icon at the bottom of the sidebar, open Model Provider and, at the bottom of the list, click Add Provider.
In Provider Name enter DaLAi, and in API Key enter your key.
In the Endpoint settings section, in both the OpenAI and Anthropic fields enter only https://dalai.ir, without /v1. The app adds it itself and shows the full path under each field.
Click Add. The list of models you can use with your key opens. Tick the models you want and click Add selected models. The app tests each model and enables the provider on its own.
On the chat screen, select the model from the DaLAi group at the top of the screen.

If a model is added to the Model List later, click the Sync models button on the same provider page. If you created the key with Model Limits, only those models appear in the list.

Not working yet

Tools that send only the Responses format, such as Codex CLI. DaLAi does not support this format.
Semantic search and creating embeddings: DaLAi does not offer an embedding model yet.

Apps we do not recommend

Some online chat services keep your key in the page URL or on their own server. A DaLAi key has access to your balance, so enter it only in an app that runs either on your own device or on a server you control.

If your tool is not on this list

It probably still works. Look for these terms in its settings:

OpenAI compatible or Custom OpenAI
Base URL or API Base or Endpoint

If there is such a field, enter the base URL and key, and take the model name from the Model List. Path details are in API overview.