Hello, everyone. I will be very appreciated if you help me to solve the problem.
It is about 8000 tokens. I ask in GPT like:
{
role: 'system',
content: "ask for check grammar and return the whole corrected text"
},
{
role: 'user',
content: "text from link about 8000 tokens",
},
I am using GPT gpt-4.1-2025-04-14, so it should return about 32000 tokens but the best result I have got was about ~7000
So almost all text corrected, but I need all text back
Help me. I know that I can use chunking but Is there will be saved context?
Additional information:
Javascript library:
const OpenAI = require(‘openai’);
Node v18.20.7
request looks like:
const { data, response: rawResponse } = await this.client.chat.completions
.create(
{
messages: this.createCheckGrammarMessages(text, dialect),
model,
temperature: this.temperature,
},
{
maxRetries: requestsRetryCount,
},
)
.withResponse();