Axel 👨💻 Developer
@axelgarciak.bsky.social
👨💻 Software Engineer 💾 Software minimalist/retro
🤖 AI tinkerer 🏗️ Building tech communities
🇪🇺 UK 🇬🇧🇩🇪🇻🇪 | Check bio: axelgarciak.com/bio
🤖 AI tinkerer 🏗️ Building tech communities
🇪🇺 UK 🇬🇧🇩🇪🇻🇪 | Check bio: axelgarciak.com/bio
Qwen3-next-80B-A3B 👀
Only 3B active parameters and almost as good (in benchmarks) as Qwen3-235B-A22B and Qwen3-32B.
Only 3B active parameters and almost as good (in benchmarks) as Qwen3-235B-A22B and Qwen3-32B.
September 11, 2025 at 7:35 PM
Qwen3-next-80B-A3B 👀
Only 3B active parameters and almost as good (in benchmarks) as Qwen3-235B-A22B and Qwen3-32B.
Only 3B active parameters and almost as good (in benchmarks) as Qwen3-235B-A22B and Qwen3-32B.
Gemma 3 270M (Million not Billion) released.
I keep a close-eye to small models, and this one is a great win.
I've seen some tests and it is clever enough despite its size, but the main purpose is to fine-tune it to do specialized tasks.
I keep a close-eye to small models, and this one is a great win.
I've seen some tests and it is clever enough despite its size, but the main purpose is to fine-tune it to do specialized tasks.
August 16, 2025 at 12:57 AM
Gemma 3 270M (Million not Billion) released.
I keep a close-eye to small models, and this one is a great win.
I've seen some tests and it is clever enough despite its size, but the main purpose is to fine-tune it to do specialized tasks.
I keep a close-eye to small models, and this one is a great win.
I've seen some tests and it is clever enough despite its size, but the main purpose is to fine-tune it to do specialized tasks.
Qwen3-coder seems to be great 👀
July 22, 2025 at 10:40 PM
Qwen3-coder seems to be great 👀
So many great announcements at Google I/O that is hard to put on a list.
I like Gemma 3n. They are decent lightweight models to run offline on smartphones.
They have 5B and 8B raw parameters but are comparable with the memory footprint of 2B to 4B models!
I like Gemma 3n. They are decent lightweight models to run offline on smartphones.
They have 5B and 8B raw parameters but are comparable with the memory footprint of 2B to 4B models!
May 20, 2025 at 8:38 PM
So many great announcements at Google I/O that is hard to put on a list.
I like Gemma 3n. They are decent lightweight models to run offline on smartphones.
They have 5B and 8B raw parameters but are comparable with the memory footprint of 2B to 4B models!
I like Gemma 3n. They are decent lightweight models to run offline on smartphones.
They have 5B and 8B raw parameters but are comparable with the memory footprint of 2B to 4B models!
Qwen3 is the ultimate GPU-Poor LLM!
Qwen3 4B and 8B are good for many use cases. Even Qwen3 0.6B seems coherent.
Qwen3-30B-A3 can be run with 4GB VRAM with enough system RAM.
I ran it on 4GB VRAM and got 12tok/s!
Qwen3 4B and 8B are good for many use cases. Even Qwen3 0.6B seems coherent.
Qwen3-30B-A3 can be run with 4GB VRAM with enough system RAM.
I ran it on 4GB VRAM and got 12tok/s!
April 30, 2025 at 11:02 PM
Qwen3 is the ultimate GPU-Poor LLM!
Qwen3 4B and 8B are good for many use cases. Even Qwen3 0.6B seems coherent.
Qwen3-30B-A3 can be run with 4GB VRAM with enough system RAM.
I ran it on 4GB VRAM and got 12tok/s!
Qwen3 4B and 8B are good for many use cases. Even Qwen3 0.6B seems coherent.
Qwen3-30B-A3 can be run with 4GB VRAM with enough system RAM.
I ran it on 4GB VRAM and got 12tok/s!
Good to see smaller LLMs pushing the pareto frontier of size vs performance.
Gemma 3 27B by Google DeepMind was released with performance between DeepSeek V3 and R1.
OlympicCoder 7B by Hugging Face with better performance than Claude 3.7 in Olympiad-Level Programming Problems.
Gemma 3 27B by Google DeepMind was released with performance between DeepSeek V3 and R1.
OlympicCoder 7B by Hugging Face with better performance than Claude 3.7 in Olympiad-Level Programming Problems.
March 12, 2025 at 7:49 PM
Good to see smaller LLMs pushing the pareto frontier of size vs performance.
Gemma 3 27B by Google DeepMind was released with performance between DeepSeek V3 and R1.
OlympicCoder 7B by Hugging Face with better performance than Claude 3.7 in Olympiad-Level Programming Problems.
Gemma 3 27B by Google DeepMind was released with performance between DeepSeek V3 and R1.
OlympicCoder 7B by Hugging Face with better performance than Claude 3.7 in Olympiad-Level Programming Problems.
One of the coolest AI use cases I've read about recently is DLSS 4 for NVIDIA GPUs.
DLSS stands for Deep Learning Super Sampling.
Games use DLSS AI to predict multiple frames and improve image quality/upscaling.
That allows lower-spec GPUs to play at higher resolutions/FPS!
DLSS stands for Deep Learning Super Sampling.
Games use DLSS AI to predict multiple frames and improve image quality/upscaling.
That allows lower-spec GPUs to play at higher resolutions/FPS!
January 13, 2025 at 9:07 PM
One of the coolest AI use cases I've read about recently is DLSS 4 for NVIDIA GPUs.
DLSS stands for Deep Learning Super Sampling.
Games use DLSS AI to predict multiple frames and improve image quality/upscaling.
That allows lower-spec GPUs to play at higher resolutions/FPS!
DLSS stands for Deep Learning Super Sampling.
Games use DLSS AI to predict multiple frames and improve image quality/upscaling.
That allows lower-spec GPUs to play at higher resolutions/FPS!
Interesting text-to-video model preserving transparency: TransPixar.
January 10, 2025 at 10:05 AM
Interesting text-to-video model preserving transparency: TransPixar.
Nvidia announced Project Digits, a personal AI supercomputer launching in May for $3K:
- Powered by GB10 Grace Blackwell Superchip.
- 128GB of unified memory.
- Can be linked to run bigger models.
It's kind of the Linux version of Mac Mini with 2x memory but 1.5x price.
- Powered by GB10 Grace Blackwell Superchip.
- 128GB of unified memory.
- Can be linked to run bigger models.
It's kind of the Linux version of Mac Mini with 2x memory but 1.5x price.
January 7, 2025 at 12:21 PM
Nvidia announced Project Digits, a personal AI supercomputer launching in May for $3K:
- Powered by GB10 Grace Blackwell Superchip.
- 128GB of unified memory.
- Can be linked to run bigger models.
It's kind of the Linux version of Mac Mini with 2x memory but 1.5x price.
- Powered by GB10 Grace Blackwell Superchip.
- 128GB of unified memory.
- Can be linked to run bigger models.
It's kind of the Linux version of Mac Mini with 2x memory but 1.5x price.
Made-up tech role #26:
Agile Stand-Up Philosopher.
Turns 15-minute meetings into existential debates.
Agile Stand-Up Philosopher.
Turns 15-minute meetings into existential debates.
January 3, 2025 at 6:38 AM
Made-up tech role #26:
Agile Stand-Up Philosopher.
Turns 15-minute meetings into existential debates.
Agile Stand-Up Philosopher.
Turns 15-minute meetings into existential debates.
DeepSeek v3 LLM is out!
According to benchmarks it is as good as GPT-4o/ Claude 3.5 Sonnet, but open source.
Here's a summary:
According to benchmarks it is as good as GPT-4o/ Claude 3.5 Sonnet, but open source.
Here's a summary:
December 27, 2024 at 8:47 AM
DeepSeek v3 LLM is out!
According to benchmarks it is as good as GPT-4o/ Claude 3.5 Sonnet, but open source.
Here's a summary:
According to benchmarks it is as good as GPT-4o/ Claude 3.5 Sonnet, but open source.
Here's a summary:
Made-up tech role #25:
Scrum Claus.
🎷 Making a list, checking it twice, gonna find out who's naughty and nice.
- Runs daily stand-ups where everyone pretends they’re on track.
- Believes every sprint should end with a Christmas miracle.
Scrum Claus.
🎷 Making a list, checking it twice, gonna find out who's naughty and nice.
- Runs daily stand-ups where everyone pretends they’re on track.
- Believes every sprint should end with a Christmas miracle.
December 25, 2024 at 7:55 PM
Made-up tech role #25:
Scrum Claus.
🎷 Making a list, checking it twice, gonna find out who's naughty and nice.
- Runs daily stand-ups where everyone pretends they’re on track.
- Believes every sprint should end with a Christmas miracle.
Scrum Claus.
🎷 Making a list, checking it twice, gonna find out who's naughty and nice.
- Runs daily stand-ups where everyone pretends they’re on track.
- Believes every sprint should end with a Christmas miracle.
Made-up tech role #24:
Silent Night SysAdmin.
Ensuring all is calm and all is bright (on the servers).
- Works tirelessly to prevent downtime during the holiday rush.
- Monitors logs while sipping eggnog and pretending everything is fine.
- Secretly wishes for a silent night without pager alerts.
Silent Night SysAdmin.
Ensuring all is calm and all is bright (on the servers).
- Works tirelessly to prevent downtime during the holiday rush.
- Monitors logs while sipping eggnog and pretending everything is fine.
- Secretly wishes for a silent night without pager alerts.
December 24, 2024 at 9:24 AM
Made-up tech role #24:
Silent Night SysAdmin.
Ensuring all is calm and all is bright (on the servers).
- Works tirelessly to prevent downtime during the holiday rush.
- Monitors logs while sipping eggnog and pretending everything is fine.
- Secretly wishes for a silent night without pager alerts.
Silent Night SysAdmin.
Ensuring all is calm and all is bright (on the servers).
- Works tirelessly to prevent downtime during the holiday rush.
- Monitors logs while sipping eggnog and pretending everything is fine.
- Secretly wishes for a silent night without pager alerts.
Made-up tech role #23:
DevOps Daredevil.
Deploying code to production on Friday afternoons, just for the thrill of it.
DevOps Daredevil.
Deploying code to production on Friday afternoons, just for the thrill of it.
December 22, 2024 at 2:06 PM
Made-up tech role #23:
DevOps Daredevil.
Deploying code to production on Friday afternoons, just for the thrill of it.
DevOps Daredevil.
Deploying code to production on Friday afternoons, just for the thrill of it.
Made-up tech role #22:
Cloud Cartographer.
Mapping the ever-shifting landscape of cloud services, one API at a time.
Cloud Cartographer.
Mapping the ever-shifting landscape of cloud services, one API at a time.
December 19, 2024 at 6:02 AM
Made-up tech role #22:
Cloud Cartographer.
Mapping the ever-shifting landscape of cloud services, one API at a time.
Cloud Cartographer.
Mapping the ever-shifting landscape of cloud services, one API at a time.
Made-up tech role #21:
Chief Procrastination Enabler.
Encouraging employees to "take a break" until nothing gets done.
Chief Procrastination Enabler.
Encouraging employees to "take a break" until nothing gets done.
December 16, 2024 at 9:19 AM
Made-up tech role #21:
Chief Procrastination Enabler.
Encouraging employees to "take a break" until nothing gets done.
Chief Procrastination Enabler.
Encouraging employees to "take a break" until nothing gets done.
Made-up tech role #20:
Chief Overthinker of Simple Problems
Turning "restart the router" into a 3-hour brainstorming session.
Chief Overthinker of Simple Problems
Turning "restart the router" into a 3-hour brainstorming session.
December 15, 2024 at 1:39 PM
Made-up tech role #20:
Chief Overthinker of Simple Problems
Turning "restart the router" into a 3-hour brainstorming session.
Chief Overthinker of Simple Problems
Turning "restart the router" into a 3-hour brainstorming session.
If you like writing/using CLI tools (command line interface), check out charm.sh
They are building useful CLI tools in golang.
They are building useful CLI tools in golang.
December 14, 2024 at 12:37 AM
If you like writing/using CLI tools (command line interface), check out charm.sh
They are building useful CLI tools in golang.
They are building useful CLI tools in golang.
He used to date a model, but...
December 13, 2024 at 9:43 AM
He used to date a model, but...
You wanted to upgrade your software version huh? 😅
December 13, 2024 at 6:26 AM
You wanted to upgrade your software version huh? 😅
Livebech, a more objective LLM benchmark, has been updated! 📊
Gemini-exp-1206 is almost at the top of the list only behind o1, well done Google!
Claude 3.5 Sonnet is still the coding king.
Llama 3.3 70b is placed between Claude3 opus and gpt-4 turbo.
Gemini-exp-1206 is almost at the top of the list only behind o1, well done Google!
Claude 3.5 Sonnet is still the coding king.
Llama 3.3 70b is placed between Claude3 opus and gpt-4 turbo.
December 8, 2024 at 3:35 AM
Livebech, a more objective LLM benchmark, has been updated! 📊
Gemini-exp-1206 is almost at the top of the list only behind o1, well done Google!
Claude 3.5 Sonnet is still the coding king.
Llama 3.3 70b is placed between Claude3 opus and gpt-4 turbo.
Gemini-exp-1206 is almost at the top of the list only behind o1, well done Google!
Claude 3.5 Sonnet is still the coding king.
Llama 3.3 70b is placed between Claude3 opus and gpt-4 turbo.
Wait for me, just one more bug to fix. 🥲
December 4, 2024 at 6:09 AM
Wait for me, just one more bug to fix. 🥲
Made-up tech role #19:
Chief Metaverse Cartographer.
Mapping out virtual worlds for people who still get lost in Google Maps.
Chief Metaverse Cartographer.
Mapping out virtual worlds for people who still get lost in Google Maps.
December 3, 2024 at 10:05 PM
Made-up tech role #19:
Chief Metaverse Cartographer.
Mapping out virtual worlds for people who still get lost in Google Maps.
Chief Metaverse Cartographer.
Mapping out virtual worlds for people who still get lost in Google Maps.
If you want to experience software minimalism:
Install Alpine Linux + DWM as your windows manager.
It's hard to get more minimal than that!
Install Alpine Linux + DWM as your windows manager.
It's hard to get more minimal than that!
December 3, 2024 at 1:19 PM
If you want to experience software minimalism:
Install Alpine Linux + DWM as your windows manager.
It's hard to get more minimal than that!
Install Alpine Linux + DWM as your windows manager.
It's hard to get more minimal than that!
Introducing: Performance Optimizer Observation Platform (Poop).
It's a tool to compare the performance of multiple commands with a colorful terminal user interface.
Stop flushing your performance down the drain.
It's a tool to compare the performance of multiple commands with a colorful terminal user interface.
Stop flushing your performance down the drain.
December 2, 2024 at 9:08 PM
Introducing: Performance Optimizer Observation Platform (Poop).
It's a tool to compare the performance of multiple commands with a colorful terminal user interface.
Stop flushing your performance down the drain.
It's a tool to compare the performance of multiple commands with a colorful terminal user interface.
Stop flushing your performance down the drain.