GPT-4 Model is a sophisticated multimodal model. Currently it accept textual inputs and producing text outputs, with the potential for image input integration in the future version GPT-5. This model outperforms its predecessors due to its enhanced general knowledge and superior reasoning abilities.
GPT-4, akin to gpt-3.5-turbo, is fine-tuned for chat applications but is equally adept at traditional completion tasks using the Chat Completions API. Dive into our GPT guide to explore GPT-4’s usage.
| LATEST MODEL | DESCRIPTION | MAX TOKENS | TRAINING DATA |
|---|---|---|---|
| gpt-4 | Surpasses GPT-3.5 models in capabilities and is chat-optimized. Scheduled for updates with the newest model version 2 weeks post its release. | 8,192 tokens | Up to Sep 2021 |
| gpt-4-0613 | A version of gpt-4 as of June 13th, 2023, inclusive of function calling data. This variant won’t undergo updates and will be phased out 3 months post the release of a successor. | 8,192 tokens | Up to Sep 2021 |
| gpt-4-32k | Mirrors the base gpt-4 model’s functions but offers 4x the context size. Slated for updates with the newest model version. | 32,768 tokens | Up to Sep 2021 |
| gpt-4-32k-0613 | A snapshot of gpt-4-32 as of June 13th, 2023. This version won’t see updates and is set for deprecation 3 months following the launch of its successor. | 32,768 tokens | Up to Sep 2021 |
While GPT-4 and GPT-3.5 models might appear comparable for rudimentary tasks, GPT-4 exhibits markedly superior performance in intricate reasoning scenarios compared to all earlier models. GPT-5 will go even further.
Read more related topics: