OpenClaw with GPT-5.4 feels noticeably better with the new update. GPT-5.4 (thinking=high + fastmode=true) seems like the sweet spot. Better judgment, less back-and-forth, still fast enough for daily use. x.com/Saboo_Shubham_…
LLMS
-

Process Reward Agents for Steering Knowledge-Intensive Reasoning
By
–
Process Reward Agents for Steering Knowledge-Intensive Reasoning paper: https://
huggingface.co/papers/2604.09
482
… -
LLMs and Small SaaS Businesses as Early AI Revolution Winners
By
–
My personal interest this year is the interaction of LLMs and small SaaS businesses. I suspect they (we?) are early big winners of the industrial revolution we’re presently going through.
-
Neurosymbolic AI Systems: Claude, AlphaGo, and Beyond
By
–
Claude Code, Alphago/AlphaEvolve/AlphaProof and Google Search, code intepreter are all neurosymbolic. Amazon, and IBM have big efforts, etc. nobody has AGI of course. but some of the best stuff out there is in fact already neurosymbolic.
-

Karpathy’s LLM Wiki: Always-On AI Agent with Memory
By
–
Karpathy's LLM Wiki idea. I built it a month ago using Gemini 3.1 Flash Lite and Google ADK. An Always-on AI agent running 24/7, continuously processing, and connecting information into memory. Same pattern. Just an LLM that reads, thinks, and writes. 100% opensource.
-
Language Models Philosophy: How AI Changes Thinking
By
–
new technology requires philosophy here’s how language models have changed how I think about the world:
-
Opus 4.7 Could Match Mythos Performance Following AI Trends
By
–
If we just had Opus 4.7 and not Mythos it might be able to do the same thing, looking at the trend.
-
1h Prompt Cache: Cost Analysis and Usage Patterns
By
–
1h prompt cache is nuanced actually. It costs more for cache writes, and less for cache reads. Whether you benefit from cheaper cache reads depends on your usage pattern — context window size, whether the query is the main agent or subagent, etc. We have been testing a
-
Prompt Cache Pricing: Read vs Write Cost Trade-offs
By
–
1h prompt cache is nuanced actually. It costs more for cache writes, and less for cache reads. Whether you benefit from cheaper cache reads depends on your usage pattern — context window size, whether the query is the main agent or subagent, etc. We have been testing a
-

Rewriting SOUL Files for GPT-5.4 Agent Migration
By
–
I rewrote every single SOUL file for my OpenClaw agents using this as a reference. If you're moving to GPT-5.4, this is a must-do. But don't just copy-paste it. Ask your agent to create a version specific to YOU based on everything it already knows about you. x.com/Saboo_Shubham_…
