Cameron Armstrong’s latest release, AmigaGPT 2.6.0, expands the capabilities of his ChatGPT client for AmigaOS and MorphOS with a range of upgrades. The application, built to work with AmigaOS 3.1 and later (including AmigaOS 4.1 and MorphOS), allows Amiga users to chat with modern OpenAI models directly from their classic systems. The update adds new models, boosts performance, and improves reliability — all while staying true to the software’s retro roots.

The latest version introduces support for GPT-4.1 mini and nano, along with the image generation model GPT Image 1. Unlike previous versions, image data is now pulled straight from OpenAI instead of being routed through a file server, noticeably speeding up the image response time. If something goes wrong, users won’t be left guessing — helpful error messages now pop up when the app encounters issues like missing MUI initialization or insufficient API credits.

For users running the M68k build, the client now runs faster thanks to new optimization flags during compilation. The update also resolves several nagging issues: image generation errors that caused crashes, speech menu toggles that took the whole app down, and that pesky bsdsocket.library error on MorphOS when quitting too early. Users who previously ran into problems after sending multiple messages should find that particular frustration is now gone as well.

The core concept of AmigaGPT remains the same — bringing AI text generation, image creation via DALL·E, and speech output to classic Amiga platforms in a way that feels native. It’s a refreshing way to keep older hardware useful, interactive, and surprisingly futuristic.

Cameron has made the source code available for those interested in digging into the project or porting it further.

For anyone still enjoying their Amiga setup, AmigaGPT 2.6.0 is worth a look — especially for developers, writers, or just the curious. It’s clever, well supported, and proves there’s still room for modern software on a classic machine.

Subscribe
Notify of
0 Comments
Oldest
Newest Most Voted
Inline Feedbacks
View all comments