Sabitlenmiş Tweet
POM
10.8K posts
POM
@peterom
Open Source AI Art hypeman w/ @banodoco | Tools for artistic expression w/ @Reigh_Art | 👫🐕 w/ @hannahsubmarine. Aspiring 2nd Renaissance Man.
Berlin/Ireland Katılım Aralık 2011
2.4K Takip Edilen15.8K Takipçiler
"Now say their limits are increasing by 50%, but don't actually give them more credits"
ClaudeDevs@ClaudeDevs
Claude Code weekly limits are increasing 50%, now through July 13. Live now for all Pro, Max, Team, and seat-based Enterprise users.
English
I'm going to write a post about how to move >80% of your token usage to open models without any compromises
This is all using Megaplan which does a lot more than just planning now considering a rebranding.
Rn I'm considering calling it:
POM@peterom
Deepseek 4 is extraordinarily capable, every time it hasn't been able to do something Opus-level, I've almost always been able to get it to do it w/ the a preparation stage $0.87/m tok for Pro may be the best deal in AI right now
English
Is it just me or did they just not do this? I was at 100% used and I’m still at 100%
ClaudeDevs@ClaudeDevs
Claude Code weekly limits are increasing 50%, now through July 13. Live now for all Pro, Max, Team, and seat-based Enterprise users.
English
@yehs688738 @ryxcommar They invest in many utopian/positive companies - BFL, Ulysses, World Labs, etc.
English
I'm not really sure what the point of this is and it's corny as all hell but what I understand even less is why everyone with a16z in their bio is having an insane crash out over it.
General Catalyst@generalcatalyst
Meet GC
English
@ryxcommar Well It's clearly attacking A16Z based on the meme that they back dystopian/cyncial companies
English
@peterom maybe but I feel like this is neither. It's just some stupid ad.
English
WOW
CODEX 5.5 WAS ABLE TO SELF DIAGNOSE THAT IT'S OWN /GOAL NEEDED AN ESCAPE HATCH
it created its own (temporary) goal to test a concept i gave it, and realized quickly that an objective in a goal it was given was impossible to reach
so it updated its own prompt (the goal) to ALLOW itself to deal with bad examples - in this case:
if it reached an example that was truly impossible to complete, it can mark that specific task as [incomplete] so it can focus on other matters related to the goal that it WAS able to solve
I'm not sure if this will cause codex to be lazy, but I'd much rather prefer if codex admitted it couldn't do something without lying (which it's already really good at!)
i've never interacted with a model this intelligent. this was 5.5 (xhigh) mode.
it feels like I'm re-discovering AI again, what an amazing model... a lot of previously impossible tasks (bc agents were too dumb to follow properly) are becoming unlocked!!!
English
@dawiddrzala They'll be able to deliver GPT5.5-level intelligence at 10% of the cost in <1 year, everyone will build their agents on top of their models and they'll have the best RL data/culture to make theirs the best
English
All OpenAI must do is not restrict Codex usage in arbitrary ways and Valhalla is theirs
ClaudeDevs@ClaudeDevs
Starting June 15, paid Claude plans can claim a dedicated monthly credit for programmatic usage. The credit covers usage of: - Claude Agent SDK - claude -p - Claude Code GitHub Actions - Third-party apps built on the Agent SDK
English
@dexhorthy Thanks for sharing this @dexhorthy, was just considering this but seems like you've done an excellent job!
English
hey surprise - you can just launch interactive in tmux and then tail the jsonl - shipped a small wrapper...ralph loop iterating to full parity rn github.com/dexhorthy/shan…
ClaudeDevs@ClaudeDevs
Starting June 15, paid Claude plans can claim a dedicated monthly credit for programmatic usage. The credit covers usage of: - Claude Agent SDK - claude -p - Claude Code GitHub Actions - Third-party apps built on the Agent SDK
English
Kimi track - Second place: sacred_not_secret by @macbethAI
An audio-reactive architectural wireframing piece where Hermes Agent sourced and optimized the temple STLs and built the custom GLSL control app with sliders and a waveform timeline.
x.com/macbethAI/stat…
MACBETH@macbethAI
sacred_not_secret My entry for the HERMES Agent Creative Hackathon by @NousResearch A thread⏬
English
The Hermes Agent Creative Hackathon sponsored by @Kimi_Moonshot has ended!
Finalists were selected by Nous and Kimi staff out of 227 submissions on creativity, usefulness and presentation.
We were absolutely blown away by the creativity of the things you all built using Hermes. It was tough choosing winners, huge thank you to all who participated!
Winners below:
English