PSA: Codex now works with Amazon Bedrock! To configure, set: model_provider = "amazon-bedrock" and Codex can run local CLI, desktop, and IDE workflows against OpenAI models on Bedrock, using AWS-native auth/IAM. Both GPT-5.4 and GPT-5.5 are supported. Big unlock for teams
TECHNOLOGY
-
Optimizing Unified Memory with Speculative Decoding on Apple Hardware
By
–
You’re optimizing the right things for Unified Memory There are some shortcomings to Speculative Decoding (especially in more creative / less coding workflows) but the pros far outweigh the cons giving the Unified Memory limitations Gonna try this on my Apple hardware tonight
-
Abacus AI Supercomputer: Build Cloud Services & Host LLMs with a Single Prompt
By
–
🚨 INTRODUCING THE ABACUS AI SUPERCOMPUTER
— Abacus.AI (@abacusai) 1 juin 2026
– literally build ANY cloud service with a prompt
– host open-source LLMs & always-on agents
– bring your own Codex, Claude or Anti-gravity
– multi-agent architectures with Opus 4.8 and GPT 5.5
– run 3D games or any other server at… pic.twitter.com/Rs8lA4G0vaINTRODUCING THE ABACUS AI SUPERCOMPUTER – Literally build ANY cloud service with a prompt
– Host open-source LLMs & always-on agents
– Bring your own Codex, Claude or Anti-gravity
– Multi-agent architectures with Opus 4.8 and GPT 5.5
– Run 3D games or any other server at -
Tools should be called from code, not from within LLM API
By
–
slowly we're all realizing that tools should be called from code, not from within the llm api
-
Anthropic restrictive, Codex is far better for code review and allows sub.
By
–
Hmmm Anthropic is very restrictive, claude -p no longer works so no. Codex is far better at code review tho and allows using the sub!
-
AI reads million posts to allow talking to them
By
–
You all sure post a lot. The AI industry, alone, posts about 30,000 times every day here on X. My AI reads it all and builds this: https://
alignednews.com/ai Now my database has about a million posts in it. So now I can "talk" to a million of your posts. Which is pretty -
New Approach to Agent Skills: Paper, Repo, and Site
By
–
Paper (arXiv):
https://arxiv.org/abs/2605.31264 (~25 min read)
Repo (GitHub, MIT):
https://github.com/titanwings/colleague-skill
… (~10 min setup)
Gallery and site:
https://titanwings.github.io/colleague-skill-site/
…
Agent Skills standard:
https://agentskills.io -
GStack by Garry Tan goes viral, 100K stars on GitHub for dev cheat code
By
–
100K GITHUB STARS IN JUST A FEW WEEKS @GarryTan
’s GStack has gone completely viral, and for good reason. The YC CEO open-sourced his personal toolkit, and it's the ultimate cheat code for devs. It turns Claude Code from a basic chatbot into an entire virtual engineering -
Perplexity AI achieves new cost-performance frontier, outperforming Anthropic
By
–
It also sets a new cost-performance frontier. On DSQA it scores 0.871, ahead of Anthropic's 0.815, at nearly half the cost per task. On WideSearch it leads on score while running cheaper.
-
Search as Code replaces tool-calling with async search primitives
By
–
The traditional tool-calling approach suffers from high latency, manual control flow, and context pollution. With Search as Code, the model composes search primitives: fanning out queries asynchronously, deduping, filtering, joining, and ranking before results hits its context.