AmigaGPT
Description:	App for chatting to ChatGPT
Download:	amigagpt.lha       (TIPS: Use the right click menu if your browser takes you back here all the time)
Size:	2Mb
Version:	2.3.0
Date:	23 Mar 2025
Author:	Cameron Armstrong (Nightfox)
Submitter:	Samir Hawamdeh
Homepage:	https://github.com/sacredbanana/AmigaGPT
Requirements:	AmigaOS 4.x + MUI + AmiSSL 5.18 or higher
Category:	network/chat
Replaces:	network/chat/amigagpt.lha
License:	Other
Distribute:	yes
Min OS Version:	4.1
FileID:	13231
Videos:
Comments:	1
Snapshots:	0
Videos:	1
Downloads:	20  (Current version)
	1219  (Accumulated)
Votes:	3 (0/0)  (30 days/7 days)

AmigaGPT is a versatile ChatGPT client for AmigaOS 3.x, 4.1 and MorphOS.
This powerful tool brings the capabilities of OpenAI?s GPT to your Amiga system,
enabling text generation, question answering, and creative exploration. AmigaGPT
can also generate stunning images using DALL-E and
includes support for speech output, making it easier than ever to interact with
AI on your Amiga. Designed to integrate seamlessly with your system, AmigaGPT
delivers modern AI technology while embracing the timeless Amiga
experience.

Features:
- Generates text based on input prompts
- Generates images based on input prompts
- UI customisation
- Full conversation history
- Text-to-speech

Requirements:
- OCS/ECS/AGA Amiga or a PowerPC machine capable of running MorphOS
- MUI 3 minimum; MUI 5 recommended for all features
- Motorola 68020 or higher CPU or PPC for AmigaOS 4/MorphOS
- For AmigaOS 3 & 4, AmiSSL 5.18 or higher
  (https://aminet.net/package/util/libs/AmiSSL-v5-OS3) for OS3 and
  (https://aminet.net/package/util/libs/AmiSSL-v5-OS4) for OS4
- Internet access using a TCP/IP stack such as Roadshow 
(http://roadshow.apc-tcp.de/index-en.php)
- MUI 3 minimum but MUI 5 recommended for all features
  (https://github.com/amiga-mui/muidev/releases)
- MCC_Guigfx MUI custom class for displaying images
  (http://aminet.net/package/dev/mui/MCC_Guigfx)
- MCC_NList MUI custom class for lists
  (http://aminet.net/package/dev/mui/MCC_NList-0.128)
- codesets.library 6.22 or higher
 (http://aminet.net/package/util/libs/codesets-6.22)
- An active OpenAI API key
- For speech on AmigaOS 3, any version of narrator.device
- For speech on AmigaOS 4, Flite Device
 (http://aminet.net/package/mus/misc/flite_device)
 - For OpenAI speech, AHI is required
 (http://aminet.net/package/driver/audio/ahiusr_4.18)

Installation:
1. Install AmiSSL and a TCP/IP stack if not already done so
2. Download and install MUI. Version 5 recommended, version 3 minimum. Reboot.
3. Download and install MCC_Guigfx, MCC_NList and codesets.library.
4. Download the latest release of AmigaGPT.
5. Extract the `amigagpt.lha` archive to a temporary location.
6. Run the provided installer.

Optional steps to enable speech functionality:
AmigaGPT supports reading the output aloud. How AmigaGPT does this depends
on whether you are using AmigaOS 3 or 4. Or for OpenAI voices, this works on
every system.

Installing AHI for OpenAI voices:

If your OS does not come with AHI installed, you can get it from
 https://aminet.net/package/driver/audio/ahiusr_4.18


AmigaOS 3:

Speech on AmigaOS 3 requires a file called `narrator.device` which cannot
bevincluded with AmigaGPT because it is still under copyright. Therefore, you
must copy this file legally from yourWorkbench disks so that AmigaGPT will be
able to synthesise speech. There are 2 versions of `narrator.device`
supported, v34 and v37. 

v34 is the original version that came with Workbench 1.x. v37 was an updated
version included with Workbench 2.0.x. It has more features and sounds more
natural, however it does sound quite different which is why AmigaGPT supports
you installing both versions and your choice of version to be used can be
selected in the Speech menu in the app.

Regardless of which version of `narrator.device` you choose to install
(or both), AmigaGPT requires that you install the free third party
`translator.library` v43. This works with both versions of `narrator.device`.

Installing `translator.library` v43:
Since `translator.library` v43 is not available as a standalone install, you
will need to install v42 and then patch it to v43.
* Download http://aminet.net/util/libs/translator42.lha and extract the
archive to any convenient location on your Amiga such as `RAM:`
* Navigate to that directory and double click the `Install` program
* Run the installer using all the default settings
* Download http://aminet.net/util/libs/Tran43pch.lha and once again extract it
 to a location of your choice
* Navigate to that directory and double click the `Install` program
* Run the installer using all the default settings
* Reboot your Amiga - It will not work until the system is restarted

Installing `narrator.device` v34:
* Insert your Workbench 1.x disk and copy `df0:devs/narrator.device` to
 `{AmigaGPTProgramDirectory}/devs/speech/34`

Installing `narrator.device` v37
* Insert your Workbench 2.0.x (you cannot use 2.1 because the speech libraries
 were removed after version 2.0.4) disk and copy `df0:devs/narrator.device` to
`{AmigaGPTProgramDirectory}/devs/speech/37`

AmigaOS 4:

Speech on AmigaOS 4 is done with the help of Flite Device.

- Download Flite Device from http://aminet.net/package/mus/misc/flite_device
- Extract the archive and run the installer

Launching AmigaGPT:
* Launch the application by double-clicking the AmigaGPT icon correspondant to
 the OS you are running
* You may also launch the app in the command line but before you do, run the
 command `STACK 32768` to give the program 32kb of stack since the default
 stack size for apps launched from the shell is 4kb and this is not enough for
 AmigaGPT and will cause random crashes due to stack overflow. This is not
 required when you launch the app by double clicking the icon since the stack
 size is saved in the icon.

Usage:
When launched, AmigaGPT presents you with a choice of opening the app in a new
screen or opening in Workbench. If you open in a new screen you have the
ability to create a screen for the app to open in. AmigaGPT supports anything
from 320x200 all the way up to 4k resolution if using a video card for RTG.
Bear in mind text will appear very tiny in resolutions above 1080p so you may
want to increase the font size settings from the MUI settings in the View menu
when the app opens.

When launching for the first time you will need to enter your OpenAI API key
before you can start chatting. If you haven't already done so, create an
OpenAI account and navigate to https://platform.openai.com/account/api-keys to
generate an API key for use with AmigaGPT.

*Chat*

When the app has opened, you are presented with a text input box. You can type
any prompt into this box and press "Send" to see the GPT-4 model's response.
The generated text appears in the box above the input. You can choose to have
this text read aloud using the "Speech" menu option. You can also select which
model for OpenAI to use in the "OpenAI" menu option.

To the left of the chat box is a conversation list which you can use to go to
another saved conversation. New conversations can be created with the "New
chat" button and conversations can be removed with the "Delete chat" button.

There are 2 main modes of operation: Chat and Image Generation. You can switch
between them via the tabs in the top left corner.

*Image Generation*

To generate images, simply select your desired image generation model from the
"OpenAI" menu then type your prompt in the text box then hit the "Create Image"
button. When it has been downloaded to your Amiga, you are then able to open
the image to your desired scale, or save a copy of the file to a new location
on your Amiga. Do note however that AmigaGPT will automatially save all your
generated images until you delete them. This is just in case you would like to
create a copy elsewhere.

*General*

In the "Edit" menu, you'll find basic text editing commands like Cut, Copy,
Paste and Clear. The "View" menu allows you to change the font used in the
chat and the UI.

The "Project" menu includes an "About" option, which displays information
about the program.

The "Connection" menu allows you to connect via a proxy server. It supports
both HTTP and HTTPS proxy servers but if you use an unecrypted HTTP proxy
server you can improve the performance of AmigaGPT by removing the need for
the encryption of the OpenAI traffic to be done on the system running AmigaGPT.
For an easy proxy server you can run on your local network you can try out
https://mitmproxy.org

Developing:
Source code hosted at https://github.com/sacredbanana/AmigaGPT
You can either compile the code natively or with the Docker container.


------
Native
------

*Building the AmigaOS 3 app*

If you would like to build this project from source you will need Bebbo's
amiga-gcc toolchain here https://github.com/bebbo/amiga-gcc

Once installed, get the required other SDK's  (AmiSSL, Translator, json-c)
from https://github.com/sacredbanana/AmigaSDK-gcc and put these in your Amiga
dev environment created in the above step.

*Building the AmigaOS 4 app*
Get this toolchain set up https://github.com/sba1/adtools

Once installed, get the required other SDK's  (AmiSSL, Translator, json-c)
from https://github.com/sacredbanana/AmigaSDK-gcc and put these in your Amiga
dev environment created in the above step.

------
Docker
------

You may use an all-in-one Docker image that is able to compile both the
AmigaOS 3 and AmigaOS 4 versions of the app.

Just install Docker on your machine and run the `build_os3.sh`,
`build_os4.sh` or `build_morphos.sh` scripts depending on which version of
the app you want to build. If you want to perform a clean build, you can set
the environment variable `CLEAN=1` for example you can run
`CLEAN=1 ./build_os3.sh`.

The build app will be saved to the `/out` directory.

Contributing:

We welcome contributions to AmigaGPT! If you have a bug to report, a
feature to suggest, or a change you'd like to make to the code, please open
a new issue or submit a pull request.

Contributors:
Code:
- Cameron Armstrong (sacredbanana/Nightfox)
 https://github.com/sacredbanana/

Art:
- Mauricio Sandoval - Icon design

Translations:
- Mauricio Sandoval - Spanish
- Tobias Baeumer - German

Special Thanks:
- Bebbo for creating the Amiga GCC toolchain https://github.com/bebbo
- OpenAI for making this all possible https://openai.com
- EAB and everyone in it for answering my questions https://eab.abime.net/
- Ján Zahurancík for all the thorough testing, bundling AmigaGPT into 
  AmiKit and for all the moral support https://www.amikit.amiga.sk
- CoffinOS for bundling AmigaGPT into CoffinOS https://getcoffin.net
- Amiga Future Magazine for reviewing AmigaGPT and publishing several of
  its updates in the News from Aminet section https://www.amigafuture.de/
- WhatIFF? Magaine for reviewing AmigaGPT and interviewing me in issue 14
  https://www.whatiff.info
- Dan Wood for reviewing AmigaGPT on his YouTube channel
  https://www.youtube.com/watch?v=-OA28r8Up5U
- Proteque-CBN for reviewing AmigaGPT on his YouTube channel
  https://www.youtube.com/watch?v=t3q8HQ6wrnw
- AmigaBill for covering AmigaGPT in the Amiga News section on his Twitch
  streams and allowing me to join his stream to promote it
	https://www.twitch.tv/amigabill
- Les Docs for making a video review and giving a tutorial on how to add
  support for the French accent https://www.youtube.com/watch?v=BV5Fq1PresE

(AmigaOS 3.2 and 4.1 only for now)
- Remove trailing newline from user messages
- Enclose all user lines in asterisks to make bold
- Fix memory leak that was present in downloading responses from OpenAI

------------------
1.3.0 (2023-11-12)
------------------
- Native PPC version for AmigaOS 4.1
- Scrollbar works again
- Improve response handling from OpenAI preventing a crash
- Latest OpenAI models added

------------------
1.2.1 (2023-11-04)
------------------
- Now fully backwards compatible with Cloanto's AmigaOS 3.X (included
 in Amiga Forever)

------------------
1.2.0 (2023-10-04)
------------------
- Now fully backwards compatible with AmigaOS 3.9

------------------
1.1.0 (2023-07-22)
------------------
- Now supports characters from Western languages other than English
- Blank responses from OpenAI no longer cause a crash

------------------
1.0.1 (2023-06-24)
------------------
- After setting UI font, the font is now applied fully without needing
  to restart the app

----------------
1.0 (2023-06-17)
----------------
- Initial release

License:

------------
Flite Device
------------

/* Copyright (c) 2013 Fredrik Wikstrom. All rights reserved.
**
** Redistribution and use in source and binary forms, with or without
** modification, are permitted provided that the following conditions
** are met:
**
** 1. Redistributions of source code must retain the above copyright
**    notice, this list of conditions and the following disclaimer.
**
** 2. Redistributions in binary form must reproduce the above copyright
**    notice, this list of conditions and the following disclaimer in the
**    documentation and/or other materials provided with the distribution.
**
** THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS `AS IS'
** AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
** IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
** ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
** LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
** CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
** SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
** INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
** CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
** ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
** POSSIBILITY OF SUCH DAMAGE.
*/