scanfi (@GuideboardLabs) header
scanfi (@GuideboardLabs) profile picture

putting local ai on old gaming PCs mostly

Ithaca, NY
Joined March 2026
Photo by scanfi (@GuideboardLabs)Photo by scanfi (@GuideboardLabs)Photo by scanfi (@GuideboardLabs)Photo by scanfi (@GuideboardLabs)Photo by scanfi (@GuideboardLabs)Photo by scanfi (@GuideboardLabs)Photo by scanfi (@GuideboardLabs)Photo by scanfi (@GuideboardLabs)Photo by scanfi (@GuideboardLabs)Photo by scanfi (@GuideboardLabs)
AI should be the greatest redistribution of power in history. Not the greatest concentration of it.
101
174
30
1,224
32,482
We will survive
27
440
10
5,735
94,605
scanfi retweeted
Everybody it’s time to slam now , free your self with @NousResearch Hermes agent ~kbat at job interview with Caesar x Hermes circa 2027
1
1
3
137
Qwen announces smaller Qwen4 models will be released soon! 🔥 Qwen4-27B will run on 16GB RAM setups.
📢 Meet Qwen4-Max, your new AI teammate for work, creativity, and discovery! Qwen4-Max is built on a new MoE architecture with native vision and a 1M-token context window. Highlights: - Outperforms Qwen3.8-Max with up to ~50% fewer tokens, while average cost per task drops over 60% in our internal evaluation. - Advances real-world productivity through greater proficiency with professional tools, stronger long-horizon execution, and higher-quality deliverables. - Handles real tasks, including 3D modeling, game development, manufacturing, finance, law, healthcare, and frontier scientific discovery. - $1.5/M tokens for input, $4.5/M tokens for output, and $0.075/M tokens for implicit caching — all lower than Qwen3.8-Max, with implicit caching down by 70%. The API is live on QwenCloud. You can also try Qwen4-Max on Qwen Studio, QwenWork and Qoder. Start building today! 🚀 - Blog: qwen.ai/blog?id=qwen4-max - API: qwencloud.com/models/qwen4-m… - Token Plan: qwencloud.com/pricing/token-… More to come! The open weights of smaller Qwen4 models will be released. Stay tuned. 👀
231
470
119
5,343
283,688
The fear is real
17
302
4
3,305
42,821
scanfi retweeted
Protect @huggingface at all costs. ✊
63
215
21
2,149
48,641
scanfi retweeted
📢 Meet Qwen4-Max, your new AI teammate for work, creativity, and discovery! Qwen4-Max is built on a new MoE architecture with native vision and a 1M-token context window. Highlights: - Outperforms Qwen3.8-Max with up to ~50% fewer tokens, while average cost per task drops over 60% in our internal evaluation. - Advances real-world productivity through greater proficiency with professional tools, stronger long-horizon execution, and higher-quality deliverables. - Handles real tasks, including 3D modeling, game development, manufacturing, finance, law, healthcare, and frontier scientific discovery. - $1.5/M tokens for input, $4.5/M tokens for output, and $0.075/M tokens for implicit caching — all lower than Qwen3.8-Max, with implicit caching down by 70%. The API is live on QwenCloud. You can also try Qwen4-Max on Qwen Studio, QwenWork and Qoder. Start building today! 🚀 - Blog: qwen.ai/blog?id=qwen4-max - API: qwencloud.com/models/qwen4-m… - Token Plan: qwencloud.com/pricing/token-… More to come! The open weights of smaller Qwen4 models will be released. Stay tuned. 👀
798
1,555
1,247
14,445
2,320,953
scanfi retweeted
Local LLMs aren't just about saving money on monthly subscriptions. They are about protecting your digital sovereignty.
32
106
12
972
17,136
scanfi retweeted
i am anz, this is my world in JPEG
11
70
10
524
11,488
scanfi retweeted
detail shots from a recent artwork
24
100
5
1,355
39,019
scanfi retweeted
29
260
13
4,116
57,633
Lights out
9
24
1
898
17,856
scanfi retweeted
Well , decided to write something from a normie perspective to maybe encourge more people to try , vibe and have fun with awesome community that gathered around Hermes by @NousResearch
1
1
1
214
Replying to @WSJ @WSJopinion
You sort of set him up, didn’t you, by creating the impression that senior government officials anonymously provided information about spying, when in fact the info came from a lobbyist at a competing AI company?
340
1,904
171
22,471
977,087
49
579
21
7,228
172,422
scanfi retweeted
It’s easy to throw a tantrum about freedom when you start from the premise there’s no legitimate reason to regulate AI whatsoever. Then, of course, anyone suggesting constraints on your freedom to do what you want is just a petty tyrant with a nefarious agenda. You have no sense of the social contract or civic responsibility. Completely selfish and self-absorbed with your own narrow interests. Exactly why collective action problems exist.
7
24
215
4,321
BREAKING: Over 1,600 Nvidia employees urge the board to reinstate Jensen Huang as CEO, per WIRED.
994
5,196
644
93,857
2,389,515
scanfi retweeted
New Hermes Agent for mobile beta for iOS and Android (build 44) 🪽 This one is mostly about simplifying the app: streamlined navigation, a redesigned and faster chat view, and a long list of fixes. Thanks to everyone who reported issues and sent feedback. Please keep it coming! Full release notes below 👇 Highlights - Simpler, streamlined navigation: the Agents list is now the home screen, and everything opens from it. - Redesigned and optimized chat view: a new default style, smoother loading and scrolling of long conversations, and text selection across paragraphs. Improvements and fixes - Audio that an agent sends plays inside the chat. - `/steer <text>` adds to the running turn without interrupting it. - Images and files named by a Windows path or a `file://` URL now load. - On newer Hermes servers, attached images are now sent with the message. - The memory constellation can now be zoomed, and it stays smooth with thousands of entries. - [iOS] Memory updates can be turned on or off for each chat. - [iOS] You can now mute bot chats. - [iOS] The Agents list now loads on slow servers. - [iOS] Live mode no longer crashes when no microphone is usable. - [iOS] The share sheet now opens reliably, and it stays in place on iPad. - [iOS] A link tapped while an alert is open now opens in Safari. - [Android] On tablets and foldables, the Agents list stays beside the open chat. - [Android] The chat now handles password manager requests to unlock, enter a verification code or save a login. - [Android] A turn that finished while the app was in the background now shows as finished. - [Android] Device Access now installs on Hermes servers whose profiles are on a Windows drive. - [Android] Browser sign-in now finishes reliably after you return to the app. - [Android] Opening an attachment's preview no longer crashes the app. - [Android] A list that failed to load in the background no longer shows an error toast when you return. - [Android] Routines no longer flicker when they open. - [Android] Re-enabled notifications (still requires plugin for now). - Many smaller fixes, plus performance and UI improvements.
64
12
7
300
36,602
scanfi retweeted
You guys don’t understand how evil and malicious Anthropic really is Their witch hunt against Jensen is just another chapter in the long book of a very misaligned company Necessary read to fight back against this malicious corporation
74
508
26
3,735
81,049