AI agents, tech, and growth | Building at https://xitter.gq/t.co/HRQ96g80hK blogging about my journey @ https://xitter.gq/t.co/mmPgJcIsV4 and pushing to https://xitter.gq/t.co/LAiwbXURmU
Portland, OR
Joined April 2021
- Tweets905
- Following324
- Followers173
- Likes1.5K
Pinned Tweet
Marimo Bot is a new project I am working on. Rough edges, still building.
It works in a folder you pick, and you make bots for the work you repeat, like writing, research or code.
site: marimobot.com/
project: github.com/DevTarlow/marimo-โฆ
blog post: tarlow.space/projects/marimoโฆ
Crunchr 1.1.0 is out. 18 tiny game sounds now, up from 11, including a new Movement group (jump, landing, hit) and a Magic group (zaps, shimmer, power-up, whoosh). All in the browser.
Use it on Itch: tarlow.itch.io/crunchr
Github: github.com/DevTarlow/crunchrโฆ
tarlow ๐ฑ retweeted
Introducing Audio8-TTS-Realistic, @Edge0_AIโs new open-weight speech model: realistic, zero-shot voice cloning in just 0.6B parameters.
โข Human-like prosody and expressive delivery
โข Clones a voice from one short clip, in 23 languages
โข 601M params: small enough to fine-tune and serve affordably
Currently #1 on the Hugging Face TTS Arena, where people vote blind on which voice sounds more human.
Open weights are available now on @huggingface. Try it free.
tarlow ๐ฑ retweeted
LFM3-1.7B just read the whole of Pride and Prejudice on a M5 Pro using MLX.
It read 100k tokens in about 28 s (sped up 8ร in the clip), then answered 100 questions back to back in 27 s, including finding a password hidden somewhere in the book.
huggingface.co/alexwengg/LFMโฆ
Introducing ๐ป๐ต๐ผ๐น: our new generation of Liquid Foundation Models. With a new tri-hybrid architecture and new multimodal encoders, the LFM3 family is the ideal model suite for specialized intelligence.
> 6 model instances, across text, audio, vision, and omni modalities, generative and decision models.
> ๐ป๐ต๐ผ๐น-๐ท.๐ฝ๐ฑ: text, 256k context window.
> ๐ป๐ต๐ผ๐น-๐
๐ป-๐ธ๐ฑ: vision-language model with video-input support for spatial understanding
> ๐ป๐ต๐ผ๐น-๐ฐ๐๐๐๐-๐ธ๐ฑ: tool calls, voice chat, reasoning in 18 languages.
> ๐ป๐ต๐ผ๐น-๐พ๐๐๐-๐ธ๐ฑ: an omni model combines vision, audio, and language.
> ๐๐ท-๐ธ๐ฑ and ๐๐ท-๐พ๐๐๐-๐ธ๐ฑ: decision models with typed outputs and calibrated probabilities in a single pass. Omni adds audio input.
> fine-tune on 1 GPU.
> a new tri-hybrid architecture
> architecture optimized for 14 processor classes across GPUs, CPUs, and NPUs.
> ๐ป๐ต๐ผ๐น-๐
๐ป-๐ธ๐ฑ is a beast. It performs competitively with Gemma 4 12B (6x larger), and Gemini 3.5 Flash Lite, and GPT-5.6 Luna, with 1.9B parameters.
> ๐ป๐ต๐ผ๐น-๐ท.๐ฝ๐ฑ is #1 on the Pareto frontier of device intelligence ๐๐๐๐๐๐๐.๐๐๐๐๐๐.๐๐
Every modality for specialized intelligence. Pick the one that fits your use case, customize it, and run it on-device.
1/ ๐งต
tarlow ๐ฑ retweeted
Introducing Gemini 4 Carbon, it's a big jump over Argon which we released just 2 weeks ago.
Gemini 4 Carbon is SOTA on cyber, long horizon tasks, ranks #1 on Artificial Analysis V5, significantly improves on usability in domains like coding, and is more aligned.
Testing LFM3-1.7B-Q8_0 on 3060 12GB
Feels good. Snappy, smart, and only using 3.4GB of the 12GB with 32k context loaded.
Current llama.cpp setup:
-c 32768 \
-fa on \
-ngl 99 \
--temp 0.1 \
--top-k 50 \
--repeat-penalty 1.1 \
I will report back next soon.
tarlow ๐ฑ retweeted
Introducing ๐ป๐ต๐ผ๐น: our new generation of Liquid Foundation Models. With a new tri-hybrid architecture and new multimodal encoders, the LFM3 family is the ideal model suite for specialized intelligence.
> 6 model instances, across text, audio, vision, and omni modalities, generative and decision models.
> ๐ป๐ต๐ผ๐น-๐ท.๐ฝ๐ฑ: text, 256k context window.
> ๐ป๐ต๐ผ๐น-๐
๐ป-๐ธ๐ฑ: vision-language model with video-input support for spatial understanding
> ๐ป๐ต๐ผ๐น-๐ฐ๐๐๐๐-๐ธ๐ฑ: tool calls, voice chat, reasoning in 18 languages.
> ๐ป๐ต๐ผ๐น-๐พ๐๐๐-๐ธ๐ฑ: an omni model combines vision, audio, and language.
> ๐๐ท-๐ธ๐ฑ and ๐๐ท-๐พ๐๐๐-๐ธ๐ฑ: decision models with typed outputs and calibrated probabilities in a single pass. Omni adds audio input.
> fine-tune on 1 GPU.
> a new tri-hybrid architecture
> architecture optimized for 14 processor classes across GPUs, CPUs, and NPUs.
> ๐ป๐ต๐ผ๐น-๐
๐ป-๐ธ๐ฑ is a beast. It performs competitively with Gemma 4 12B (6x larger), and Gemini 3.5 Flash Lite, and GPT-5.6 Luna, with 1.9B parameters.
> ๐ป๐ต๐ผ๐น-๐ท.๐ฝ๐ฑ is #1 on the Pareto frontier of device intelligence ๐๐๐๐๐๐๐.๐๐๐๐๐๐.๐๐
Every modality for specialized intelligence. Pick the one that fits your use case, customize it, and run it on-device.
1/ ๐งต
tarlow ๐ฑ retweeted
๐ขGIVEAWAY๐ข
As a way of saying thank you for being rad, one lucky winner will get it all! ๐ฅ๐ฑ
To enter: FOLLOW, LIKE, & RTโ๏ธ
One person takes it all:
๐น๏ธAtari Console with 130+ games
๐น๏ธTomagotchi in box
๐น๏ธ2 Nicktoon toys
๐น๏ธComplete in box Annalog NES
Ends 10/30๐ฅ
Do you like learning about AI?
Build your AI skills with IBMย SkillsBuild learning catalog.
skillsbuild.org/learning-catโฆ
If you are getting into AI or already locked in.
@IBM breaks down terms and concepts in easy to understand ways.
I always learn something there.
youtube.com/@IBMTechnology
When you are not coding what are you doing?
I like tending to my plants. ๐ฑ
Right now I have over 30 in my studio, it's starting to look like a jungle and I love it.
And yes, most of them already have names ๐
Are you into?
- AI
- DeepSeek
- Agents
- Automation
- Building
- Learning
- Growth
- Helping
Then what are you waiting for?
You are my people.
Let's connect and start growing. ๐ฑ
Yeah same here @NoemiTitarenco , I gave them another shot when they adjusted usage, but I cancelled them the other day too.
Surely there more concrete evidence? I know @paulgug is doing nerf watch, but do we need a constant nerf watch on subs?
My main grime with Opencode is it does drop out quite a bit which can interrupt my workflow - but I've never done a side by side comparison to see.
