FancyRocket

Macbook Pro M5 Max 128GB | LM Studio | Pi Coding Agent | OpenCode | VSCode | C# | MonoGame |

[FancyRocket @ldreisi]

@MordHaze @Arride96 Looks good bro. Work on the Grass, Dirt, Stone next!

[Jun Song @jun_song]

Next prediction : we will be able to run Opus-4.8 level local AI on a Macbook in 3 months.

My account's prediction success rate is almost 100% so far.

[FancyRocket @ldreisi]

@TencentHunyuan Will there be smaller models <100b?

[FancyRocket @ldreisi]

@ItsmeAjayKV Fable 5 Intelligence on a single 5090 when? Haha

Seriously, local llms are cooking!

[AJ @ItsmeAjayKV]

It's crazy to think that most of my tasks can be done with local models now.

My current local model stack for @NousResearch Hermes on my single 3090:

Gemma4-12B-QAT (KV F16 + full context)-> normal everyday tasks. quick research.

Qwen3.6-35B-A3B-Q6/Q8 (KV q8_0 + 144k context + offloading) -> serious agentic tasks + coding tasks

Qwen3.6-27B-Q4/Q5 (KV q8_0 + 110k context) -> only for coding tasks

All running through llama.cpp.

[FancyRocket @ldreisi]

@MordHaze @Arride It's nice to have someone help and make the pixel art. The coding is already a handful haha

[FancyRocket @ldreisi]

@Youssofal_ Awesome, thanks! I am excited to try this out

[FancyRocket @ldreisi]

@Youssofal_ Will the update be available through V1 or will I have to re-download

[kuch (vibecoding arc) @thekuchh]

HE FED HIS SECOND BRAIN 1,500 CONVERSATIONS AND IT STARTED THINKING FOR HIM

most second brains just remember. a notes app stores what you wrote and never has a thought of its own. this one thinks

this one reads your actual history, 1,391 Claude conversations, 138 Claude Code sessions, 238 articles, and compiles it into a living wiki, one page per topic, that gets smarter every run

then four agents read you from four angles:

> Post: finds what actually made you money, with live proof from the web
> Build: surfaces what you started but never shipped, tells you what to make next
> Stoic: reads your journal and psychology, coaches you through the day
> Note: studies your best posts, hands you the next one to write

every 6 hours they update the wiki and write you something new. while you sleep

the whole build, every prompt and file:
[article link]

bookmark this

[FancyRocket @ldreisi]

@jun_song Hopefully Mini models replace or are equivalent to GLM 5.2 soon. Think this will happen?

[FancyRocket @ldreisi]

@mertunsal2020 Will you be releasing a coding model that is similar in size?

[FancyRocket @ldreisi]

@sophiamyang New coding model when? 🤞🙏

[MordHaze @MordHaze]

The world of #EchoesOfNawia is starting to flow! 🌊✨

Water generation & spreading systems are alive. Seeing rivers shape the map while waiting for @Arride96's textures is a huge milestone for our dark fantasy vibe.

What’s your favorite #RPG water mechanic? #IndieDev #GameDev

[FancyRocket @ldreisi]

@thsottiaux @bridgemindai GPT OSS 120B v2 soon? 😅

[FancyRocket @ldreisi]

@MordHaze Yes, that's right. Pooling and Caching are huge, Dirty lighting is a good method for lighting. Definitely stay away from LINQs

Update() is where a lot of the work happens so only have what is necessary

Just have Draw() submit sprites/text through something like SpriteBatch

[FancyRocket @ldreisi]

@MordHaze Multihreading is a pain, i tried that before and went back to single thread after implementing liquids, lighting, multiplayer into my game. I kept running into race conditions and deadlocks.

ALSO, use GIT and make commits often, will save your life someday lol

[FancyRocket @ldreisi]

@MordHaze No problem! I have done a lot of this before. I definitely had to rewrite systems before because I was an idiot lol

The two biggest challenges will be lighting and multiplayer it took me over a year to get my lighting system looking good and well optimized 💀

[FancyRocket @ldreisi]

@MordHaze I think you could make MonoGame work, its basically the same framework as Terraria and that game has a crazy amount going on. You have to remember they have been adding on and optimizing it for over a decade