From Proprietary Chains to Free Software: How I Generated 18 Stories in 160 Seconds Using My Own Hardware
Description: In this article, I show you a real-world example of how running free software on your own hardware gives you complete control over your digital creativity. I generated 18 long stories simultaneously using six instances of a large language model on a single GPU. No cloud. No tracking. No “content policy” blocking my ideas. Just freedom.
I woke up this morning wanting to test something practical. I have a computer with a GPU. I have free software. I wanted to see if I could push my machine to produce creative writing at scale, entirely under my control.
So I did something that most people using proprietary AI services would never dream of. I ran six instances of a large language model on a single GPU. Each instance had four concurrent processing slots. That gave me twenty-four potential simultaneous conversations. I decided to test it with eighteen long stories.
Here is what happened. I wrote a tiny script that sent three requests to each of my six servers. Each request asked for a fifteen hundred word story about artificial intelligence. Nothing special. Just a creative writing task.
I pressed enter. Then I watched.
Within 160 seconds, all eighteen stories had completed. The fastest took 135 seconds. The slowest took 160 seconds. The stories ranged from 1377 words to 1749 words. That is nearly thirty thousand words generated in under three minutes. On a single consumer GPU. Using free software.
Let me show you the numbers because numbers do not lie:
๐ Starting 18 async story generations (6 ports x 3 requests)...
๐ Results: /tmp/stories_20260507_115208
curl -s -X POST http://127.0.0.1:1234/v1/chat/completions \
-H "Content-Type: application/json" \
-d '{"messages":[{"role":"user","content":"Say OK"}],"max_tokens":5}' \
| jq '.choices[0].message.content'
โ
Port 1239, Request 2: 135023ms, 1377 words
โ
Port 1237, Request 2: 136317ms, 1417 words
โ
Port 1236, Request 1: 139545ms, 1429 words
โ
Port 1234, Request 1: 144551ms, 1492 words
โ
Port 1235, Request 2: 145963ms, 1484 words
โ
Port 1234, Request 2: 149021ms, 1496 words
โ
Port 1238, Request 1: 149063ms, 1555 words
โ
Port 1238, Request 3: 154738ms, 1602 words
โ
Port 1238, Request 2: 155557ms, 1585 words
โ
Port 1234, Request 3: 155818ms, 1595 words
โ
Port 1236, Request 2: 157813ms, 1645 words
โ
Port 1237, Request 1: 157824ms, 1638 words
โ
Port 1239, Request 3: 158371ms, 1618 words
โ
Port 1237, Request 3: 158525ms, 1633 words
โ
Port 1239, Request 1: 158783ms, 1688 words
โ
Port 1235, Request 1: 158904ms, 1638 words
โ
Port 1235, Request 3: 159685ms, 1693 words
โ
Port 1236, Request 3: 160048ms, 1749 words
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
โ
All 18 stories completed!
๐ Results saved to: /tmp/stories_20260507_115208
Quick stats:
Avg duration: 2ms
Total stories: 18
Port 1239, Request 2: 1377 words in 135 seconds
Port 1237, Request 2: 1417 words in 136 seconds
Port 1236, Request 1: 1429 words in 139 seconds
Port 1234, Request 1: 1492 words in 144 seconds
Port 1235, Request 2: 1484 words in 145 seconds
Port 1234, Request 2: 1496 words in 149 seconds
Port 1238, Request 1: 1555 words in 149 seconds
Port 1238, Request 3: 1602 words in 154 seconds
Port 1238, Request 2: 1585 words in 155 seconds
Port 1234, Request 3: 1595 words in 155 seconds
Port 1236, Request 2: 1645 words in 157 seconds
Port 1237, Request 1: 1638 words in 157 seconds
Port 1239, Request 3: 1618 words in 158 seconds
Port 1237, Request 3: 1633 words in 158 seconds
Port 1239, Request 1: 1688 words in 158 seconds
Port 1235, Request 1: 1638 words in 158 seconds
Port 1235, Request 3: 1693 words in 159 seconds
Port 1236, Request 3: 1749 words in 160 seconds
Every single story saved to my hard drive. Every single word generated under my roof. No company looking over my shoulder. No “safety filters” deciding what I could write. No usage limits. No monthly fees.
This is what free software gives you. Not just the freedom to run the program for any purpose. But the practical, tangible ability to scale your creativity without asking permission.
Here is the key point that the proprietary cloud services do not want you to understand. When you run software on your own hardware, you own the capacity. You decide how many instances to run. You decide what prompts to send. You decide what to do with the output.
The proprietary services want you to believe that you need their infrastructure. They want you to pay per token. They want you to accept their content policies. They want you to trust them with your ideas.
But I just showed you that a single GPU can generate nearly thirty thousand words in under three minutes. My GPU is nothing special. It is a consumer card. You can buy one. You can install free software on your computer today and do exactly what I did.
The software stack I used is entirely free. The model weights are freely distributed. The server software comes from the free software community. The operating system is GNU/Linux. Every single component respects your freedom.
This matters because when you rely on proprietary AI services, you lose control. The service can change its terms tomorrow. It can start censoring your prompts. It can raise prices. It can shut down. It can read your conversations. It can train its models on your data. You have no recourse. You are a user, not an owner.
When you run free software on your own hardware, you are the owner. You control the machine. You control the data. You control the output. No one can take that away from you.
I invite you to read about what free software actually means. The GNU Project has been explaining this since 1983. The philosophy is simple: software should respect your freedom. The famous definition at https://www.gnu.org/philosophy/free-sw.html spells it out clearly. You have the freedom to run the program as you wish. The freedom to study and change it. The freedom to redistribute copies. The freedom to distribute your modified versions.
These freedoms are not abstract. They are practical. They let me run six copies of the same model on one machine. They let me write scripts that generate eighteen stories simultaneously. They let me push my hardware to its limits without asking anyone for permission.
If you want to participate in this world, come join us. The free software community is active and welcoming. You can find mailing lists, chat rooms, and forums where people share knowledge and help each other. Start at https://lists.gnu.org/ to find projects that interest you.
Do not let the proprietary cloud vendors lock you in. Your computer is yours. Your creativity is yours. Free software gives you the tools to exercise that ownership.
I ran eighteen long stories in one hundred sixty seconds. You can too. The only difference is whether you choose freedom or you choose a cage.
Jean Louis, Free Software Supporter since 1999