Rohan De Silva

AI Engineer focused on LLMs, agents, and workflow automation. Founder at TrelviqAI.

[Rohan De Silva @rohandesilvaa]

Liquid AI released LFM2.5-2.6B, an agentic model.

That runs fully on-device. It is small, but capable to power agentic workflows: planning, calling tools, and tackling multi-step tasks.

> Pre-trained on ~34T tokens
> LFM2.5 flagship hybrid architecture
> Context length: 128K

[Rohan De Silva @rohandesilvaa]

Today Qwen 3.8-Max released!

The most capable model in the Qwen family.

It scales to 2.4 trillion parameters, delivering comprehensive improvements across coding, work, research, and long-horizon tasks. 

This did over a 10+ day long-horizon autonomous coding run.

[Rohan De Silva @rohandesilvaa]

We are close to singularity.

OpenAI internal version of Astra, solved 10 long-standing open problems in mathematics and theoretical computer science.

That model is their next major model. (will be GPT 6)

https://t.co/E0cTJMVFiB

[Rohan De Silva @rohandesilvaa]

@muratworks Report

[Rohan De Silva @rohandesilvaa]

@s_chiriac Yeah. I agree 🙌

[Rohan De Silva @rohandesilvaa]

@s_chiriac Remote 🙌

[Rohan De Silva @rohandesilvaa]

@matteo_spada Wel come Matt. 🥰

[Rohan De Silva @rohandesilvaa]

@yu_angela Congratulations 🎉

[Rohan De Silva @rohandesilvaa]

@varshadr1234 C++