Joined November 2023
- Tweets 296
- Following 134
- Followers 334
- Likes 208
61 Photos and videos
Pinned Tweet
Mar 2
PlebDevs Adversarial AI course is ready to start!
Lesson 1 is now released and live on the plebdevs platform!
Rest of the course will be released over the coming weeks and yes it is all totally FREE
1
3
15
5,297
plebdevs retweeted
Apr 20
Trying kk2.6 - thoughts so far:
- Incredible at spawning / managing subagents
- Great at codebase exploration / bug hunting
- Good thinking / planning model
- Much better at tool calling (then prev kimi models)
- Solid at long running tasks
I don't believe the benches but the vibe check is solid so far
Apr 20
Meet Kimi K2.6: Advancing Open-Source Coding
🔹Open-source SOTA on HLE w/ tools (54.0), SWE-Bench Pro (58.6), SWE-bench Multilingual (76.7), BrowseComp (83.2), Toolathlon (50.0), Charxiv w/ python(86.7), Math Vision w/ python (93.2)
What's new:
🔹Long-horizon coding - 4,000 tool calls, over 12 hours of continuous execution, with generalization across languages (Rust, Go, Python) and tasks (frontend, devops, perf optimization).
🔹Motion-rich frontend - Videos in hero sections, WebGL shaders, GSAP Framer Motion, Three.js 3D.
🔹Agent Swarms, elevated - 300 parallel sub-agents × 4,000 steps per run (up from K2.5's 100 / 1,500). One prompt, 100 files.
🔹Proactive Agents - K2.6 model powers OpenClaw, Hermes Agent, etc for 24/7 autonomous ops.
🔹Claw Groups (research preview) - bring your own agents, command your friends', bots & humans in the loop.
-
K2.6 is now live on kimi.com in chat mode and agent mode.
For production-grade coding, pair K2.6 with Kimi Code: kimi.com/code
-
🔗 API: platform.moonshot.ai
🔗 Tech blog: kimi.com/blog/kimi-k2-6
🔗 Weights & code: huggingface.co/moonshotai/Ki…
1
3
230
Apr 20
gpt-oss:20b is still maybe the BEST all around small model
(as far as quality, size, speed, breadth)
gemma4:26b arguably better but only time will tell. gpt-oss has been SOLID for like a year at this point though, it has LEGS!
1
1
68
Apr 13
Few understand how far local AI has come over the last 6 months...
Apr 12
Carnice 9B and TurboQuant on RTX 3060 Ti 8GB
(my 10 y old gaming PC)
Absolutely insane...
A 9B model fits poorly event at 4-bit on this setup, normally would choke at 32K context but turboquant stretches to 131K (still with ~56tps) and allows me to run full autonomous Hermes agent setup locally
Completely FREE and opensource
1
2
102
Apr 10
Have you checked out llmfit?
Very dope tool for easily finding out what local models you can run on your machine
github.com/AlexsJones/llmfit
7
67
Apr 7
My hermes agent (running gemma4:26b locally btw) used the new manim-video skill to make a video diagram describing how hermes works - not bad!
5
202
Apr 3
Ok yeah this is huge...
Ollama is now updated to run the fastest on Apple silicon, powered by MLX, Apple's machine learning framework.
This change unlocks much faster performance to accelerate demanding work on macOS:
- Personal assistants like OpenClaw
- Coding agents like Claude Code, OpenCode, or Codex
2
103
plebdevs retweeted
Apr 2
Meet Gemma 4!
Purpose-built for advanced reasoning and agentic workflows on the hardware you own, and released under an Apache 2.0 license.
We listened to invaluable community feedback in developing these models. Here is what makes Gemma 4 our most capable open models yet: 👇
167
837
7,154
626,790
Mar 31
New PlebDevs AI Guide: Goose Ollama
A full local agent system with two easy opensource tools 🤝
1
1
9
283
Mar 30
Ollama already made it dead simple to find, pull, and run local models. Now they're shipping the best open coding agent on top of that 👀
Runtime model directory agent, all in one tool. Local agentic workflows are getting so much more accessible
ollama launch pi --model kimi-k2.5:cloud
Ollama can now launch Pi, the coding agent that powers OpenClaw.
Designed to be a minimal coding harness that can be adapted to your workflows to create your own coding agent.
Comes bundled with powerful primitives to build on, and can be extended with extensions, skills, prompt templates, and themes.
All Pi packages work with Pi & Ollama, making it infinitely customizable for different tasks and use cases.
1
1
6
371
Mar 29
Scrounge your GPU’s into a super rig, ape into that Mac Mini, go into debt for the DGX Spark DO what is needed to run increasingly more and larger open source AI models.
THIS is our shitcoin 🤝
90
Mar 26
Latest lesson - TECHSTACK
4
83
Mar 25
Running @OsaurusAI - Mac-native local AI suite that makes MLX models as easy to run as Ollama, but with a real UI, persistent memory, sandboxed code execution, and 20 plugins out of the box. Ollama API compatible too. If you're on Apple Silicon, worth a look 👀
2
5
251