Mihai Constantin

[Malvin @ManOfFocus_]

BREAKING NEWS: Neverrrrrrrr... Neverrrrrrrr... Neverrrrrrrrrrrrrrrrrrrr look desperate in life. Remain calm during hard times. Understand it's just your time to suffer like every great man before you. You're a man.

[Mihai Constantin @CoMiK6891]

Nothing new, but you will truly know the value of a person when you lose them. And then there’s the question: Could I have done more to help them? If the answer is “Yes,” then there’s a lesson, albeit a bitter one.

[Mihai Constantin @CoMiK6891]

They said "just start posting". So i am.

[Mihai Constantin @CoMiK6891]

@ciruai Losing gear / equipment is always an opportunity to get a newer / better one.

[Mihai Constantin @CoMiK6891]

@StefanRows hi ! senior web dev here. skateboarding with your son, that is 100% awesome.

[Mihai Constantin @CoMiK6891]

@pupposandro @luceboxai you italians rock at espresso and AI / open LLM

[Mihai Constantin @CoMiK6891]

@varien seriously cool

[Mihai Constantin @CoMiK6891]

@ishu_codes great stuff! liked and followed, keep it up!

[Mihai Constantin @CoMiK6891]

@techiester83 @henrikklee @shayne_legault @dhh @omarchy this is seriously cool

[Brian @bkuyper]

Inference engine built in rust currently targeting AMD R9700 GPU's.

@AIatAMD #AMD #R9700 #LocalAI #LocalLLM

https://t.co/QlJ8MJOpMJ

[Kimi.ai @Kimi_Moonshot]

Releasing the model weights of Kimi K3.1 Pro: https://t.co/tJXcYUeSCj
We'll keep advancing open intelligence.

[bebu @bebuu_ai]

https://t.co/UStDmf5MNx chief scientist Jie Tang just revealed the training pipeline behind the upcoming GLM-5.5 (32B, 122B, and 256B) during a lecture at Tsinghua University today.

Instead of relying on public datasets, their pretraining pipeline pulls from six separate internal engines:

- Web: 1 trillion pages crawled at ~10B pages/day with PB-scale template clustering
- Documents: 2 billion scientific PDFs parsed through GLM-OCR (4x faster with speculative decoding)
- Books: 50 million books with layout reordering and garbled-text filtering
- Code: 500 million repositories turned into executable tasks with real tests and issue descriptions
- Video: 10 billion videos for long-sequence world knowledge
- Audio: 300 million hours rewritten via ASR into trillions of tokens

[Mihai Constantin @CoMiK6891]

@noisyb0y1 Yeah, Grok Bot @bot is really cool.

[Mihai Constantin @CoMiK6891]

@oanaolt Wow, nu ma asteptam la un raspu s atat de rapid ! Tocmai ma amuzam copios despre asta https://t.co/rxv1dOj2xW :))))

[Oana Olteanu @oanaolt]

What’s like to interview rn

[Mihai Constantin @CoMiK6891]

@tonysimons been using hermes for a while now as high level manager, the agent i actually work with directly - it then calls on opencode agents as it needs, according to an opencode skill, being aware of the agents opencode has - and it delivers

[Mihai Constantin @CoMiK6891]

@SamJWasserman @u1tra_instinct AMD Strix Halo support would be highly appreciated, if possible. The AMD / Radeon ecosystem & community would greatly benefit from it.

[Mihai Constantin @CoMiK6891]

@TeksEdge also you end up way better with 2x strix halo 128gb price-wise and memory-wise.

[Mihai Constantin @CoMiK6891]

@TeksEdge the official specs were clear that there is little to no performance increase compared to strix halo. only ram size is increased. and the price of course :))

[MiniMax (official) @MiniMax_AI]

MiniMax M3.1 Flash: Space Performance + Bunny Speed

🚀🐰 Yes, Space Bunny - which has topped global leaderboards with 200T+ tokens - was an earlier version of M3.1 Flash.

💰Redefining the performance economics of agentic AI. M3.1 Flash achieves frontier-level intelligence across key tasks such as Cowork (Automation-Bench: 62%; ALE: 30%), Coding (DeepSWE: 74%) and Visual Reasoning (BabyVision: 91%), at 1/3 the cost of M3 and 3x the speed.

API pricing (flat rate): $0.15 / $0.60 per 1M input/output tokens, with cache reads at $0.015 / 1M tokens.

Model weights will arrive on Oct 23.

API: https://t.co/fHRdSV73Hr
M Plan: https://t.co/XfJWSyauda
MiniMax Code for desktop: https://t.co/SGmLTI2HQm