Configuring OctoBot to Use ChatGPT and Other LLMs for Trading
Summary
This guide explains how to connect OctoBot’s GPT interface to a language model for trading evaluations. For OpenAI, it describes adding an API key in the interface configuration, enabling the GPTEvaluator, and choosing a model through evaluator settings. It also explains that a custom base URL can point OctoBot to another compatible LLM service, including a local Ollama server.
The cost section notes that API requests consume tokens and suggests estimating usage from the number of evaluator calls. Its example links call frequency to the selected trading timeframe, number of pairs, and exchanges. The page does not describe how the model produces trade predictions, assess signal quality, or report backtest or live results. It therefore serves as integration and cost guidance, not evidence that LLM-generated decisions are profitable or reliable. Costs and model behavior may depend on the provider and configuration.
Key ideas
- OctoBot can connect to an OpenAI model through its GPT interface and an API key.
- A custom base URL allows connection to other compatible language model services.
- Local Ollama use is configured by pointing the base URL to the Ollama server.
- The GPTEvaluator settings control which model is selected for trading evaluations.
- API usage costs depend on request frequency and token consumption, while trading performance is not evaluated.
Tags
Full text
# Trader avec ChatGPT
---
title: "ChatGPT"
description: "Apprenez comment configurer votre OctoBot pour trader en utilisant l'IA et ChatGPT ou d'un autre LLM. Comprenez les coûts d'une configuration locale."
sidebar_position: 3
---
# Trader avec ChatGPT
:::info
La traduction française de cette page est en cours.
:::
Seamlessly [Integrate ChatGPT within your trading strategies](/guides/octobot-trading-modes/chatgpt-trading) and profit from the power of AI trading.
<div style={{textAlign: "center"}}>
</div>
Checkout the [ChatGPT trading guide](/guides/octobot-trading-modes/chatgpt-trading) to learn more about how to trade with ChatGPT using OctoBot
OctoBot uses the ChatGPT interface to interact with ChatGPT.
## Configuration du service ChatGPT
To use ChatGPT on an open source OctoBot, the only configuration you need is to enter your OpenAI API key into the GPT Interface
1. Create or login to your <a href="https://platform.openai.com/" rel="nofollow">OpenAI</a> account
2. Create a new API key on <a href="https://platform.openai.com/account/api-keys" rel="nofollow">your account settings</a>
3. In the Accounts tab of the web interface, add the `GPT` interface if missing
4. Copy your API key into the `openai-secret-key` GPT configuration
## Prédictions avec un LLM personnalisé
OctoBot can connect to any LLM using the **LLM custom base url** configuration parameter. This is useful to use other AI models than the default OpenAI ones.
In this case, the **Secret key** parameter, will be used to authenticate to this other LLM server when necessary. It will be ignored otherwise.
## Trader avec des prédictions Ollama
To connect to a local Ollama LLM model, configure the **LLM custom base url** of your OctoBot to your Ollama server address followed by `/v1`.
Using the default Ollama address (`localhost:11434`), your **LLM custom base url** would then be: **`http://localhost:11434/v1`**.
## Sélectionner votre modèle de LLM
Selection of the LLM model to use is configured in your GPTEvaluator. When your GPT interface as configured and your `GPTEvaluator` is enabled (when using a ChatGPT-based profile or a custom profile using the `GPTEvaluator`), you can select the LLM model to use from your GPTEvaluator configuration.
The `GPTEvaluator` configuration interface can be accessed from your profile or directly from the `/config_tentacle?name=GPTEvaluator` path of your OctoBot web interface.
## Coûts
Using ChatGPT from automated API calls is a paid service from OpenAI. Each call to ChatGPT will consume
a few OpenAI tokens.
Each call to ChatGPT is recrating a request which usually consumes around 90 OpenAI tokens.
You can get the current price of OpenAI token from <a href="https://openai.com/pricing" rel="nofollow">the OpenAI pricing page</a>.
You can estimate the cost of using ChatGPT related features by estimating the amount of requests per day.
> Running a strategy on 4h for 2 trading pairs on 1 exchange: the GPT evaluator will be called every
> 4 hours for each trading pair for each exchange.Shown in full with attribution under the source's licence. Licence: GPL-3.0
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.