Guide

GPT-4 Turbo

GPT-4 Turbo is the model that made 128K-token context mainstream. On Magnus it is the veteran of the OpenAI lineup - still a strong writer with a distinct, careful style some long-time GPT users prefer.

← All guides

At a glance

Announced by OpenAI in November 2023 and promoted to general availability on April 9, 2024. It brought the 128K context window, fresher training data, and vision support to the GPT-4 line at a third of the original GPT-4's price.

On the All-inclusive plan a GPT-4 Turbo reply costs 5 credits - premium-tier pricing that reflects its API cost, not its ranking; the newer GPT-4o outperforms it for less.

Benchmarks

  • MMLU: 86.5%.
  • 128K-token context window - roughly 300 pages of text in a single conversation.

Best at on Magnus

Long pasted documents and users who know and like the classic GPT-4 voice. For most new setups, GPT-4o (3 credits) or GPT-5.6 Luna (1 credit) are the better value.