Ian Greenberg10

[Ian Greenberg10 @iangreenberg1]

@deepseek_ai Please drop the weights 🙏

[Ian Greenberg10 @iangreenberg1]

@preem1nent @Blackwellboy @MiaAI_lab Lagunsa s 2.1, qwen 3.6 27b, Ornith 35b

[Ian Greenberg10 @iangreenberg1]

@Tech2Wild Real

[andi (twocents.com) @Nexuist]

Never give up. Never ever give up

[Ian Greenberg10 @iangreenberg1]

@LottoLabs Have you gotten good results with 27b in omp I wanted to love it but felt it was to bloated for the model. I ended up getting much better results in pi.

[Ian Greenberg10 @iangreenberg1]

@sudoingX I find hermes pretty bloated for local agents (qwen 3.6 27b and down)

[Ian Greenberg10 @iangreenberg1]

@blackwellboy @UnslothAI So youre saying I can run Glm 5.2 gguf 2bit 👀

[Ian Greenberg10 @iangreenberg1]

@Tech2Wild Gemma 12 + miso tts and setup a voice assistant to control 27b

[Ian Greenberg10 @iangreenberg1]

@Youssofal_ I highly recommend using opencode with https://t.co/dRNQ7iIKdS I have a minimax 2.7 reap orchestrating/reviewing qwen 3.6 27b (qwopus) and it has blown away my expectations. Also I have found Jcode works well with local models https://t.co/gEAfoKhPIj

[Ian Greenberg10 @iangreenberg1]

@bridgemindai Highly recommend minimaz 2.7 reap smarter then qwen 3.6 and easily runs on that setup. Could even run minimax 2.7 reap on one and qwen 3.6 a3b/27b on the other one and have them work together

[Claude @claudeai]

Introducing Claude Opus 4.6. Our smartest model got an upgrade.

Opus 4.6 plans more carefully, sustains agentic tasks for longer, operates reliably in massive codebases, and catches its own mistakes.

It’s also our first Opus-class model with 1M token context in beta.

[Markov @MarkovMagnifico]

how my codebase written entirely with claude code runs

[dei @parcadei]

https://t.co/0XBD6OEvtP

continuous claude v3 is finally done

same problem, new architecture, better engineering.

a setup designed to explore solutions to recurring obstacles in agentic coding:

context, memory, learning, math, codebase exploration and much more

strap in, we're about to go speed run the new setup ↓

[Sean VanWinkle @SeanV6790]

1/ I just open-sourced HMLR — the first memory system that passes every impossible test at 1.00/1.00 on gpt-4.1-mini.

No 128k context. <4k tokens average.

https://t.co/hOjVsB6UBm

@elonmusk @karpathy @langchain @LlamaIndex @hwchase17 @andrew_ng @gwern @repl_it

[Ian Greenberg10 @iangreenberg1]

@SimpOnSolanaa G3xwjVnqinQCAwfeiLGVJCQknED2KgBQ7oCyEsynwdaK

[Ian Greenberg10 @iangreenberg1]

@Trmpsolana G3xwjVnqinQCAwfeiLGVJCQknED2KgBQ7oCyEsynwdaK

[Ian Greenberg10 @iangreenberg1]

@PeterSolanaa G3xwjVnqinQCAwfeiLGVJCQknED2KgBQ7oCyEsynwdaK

[EmProps @emprops_ai]

@iragreenberg 's historic 10K LFG PFP collection "Avatars of Goa" launches in 1 hour. Bulk minting up to 10 at a time, 1 $xtz each. Only on @emprops_io. #LFG_AiArt #Oracles_Sequel #duh