26 October 2026
Blog says OpenAI and Anthropic cut LLM prices by up to 80%
Falling model prices lead the day. HighLevel shipped several updates for agencies, mostly around reviews and social media. An Apollo economist also says AI has paid off little for firms outside tech.
Models
OpenAI and Anthropic cut LLM prices by up to 80% (opens the source in a new tab)
A blog post says OpenAI and Anthropic cut their LLM prices by up to 80% within a few months. The post asks why. The claim comes from the blog, not from the companies.
Why it matters: If you build on these models, check your current prices and see whether your costs can drop.
minimaxir.com via Hacker News · 25 Oct
Cerebras deploys trillion-parameter AI models (opens the source in a new tab)
Cerebras says it has deployed trillion-parameter AI models. The feed gives no further detail.
Why it matters: Watch whether this shows up as faster or cheaper options in the tools you already pay for.
cerebras.ai via Hacker News · 25 Oct
Microsoft Decision-1, a 9B Qwen model, explained (opens the source in a new tab)
A post explains the launch of Microsoft Decision-1, described as a 9B Qwen model. This comes from a third-party blog, not from Microsoft.
Why it matters: Small models can be cheaper to run, so look for Microsoft's own announcement before you plan around it.
shattered.io via Hacker News · 26 Oct
Agents
HighLevel Reviews AI Agents get more controls (opens the source in a new tab)
HighLevel added Brand Voice controls, Quality Guardrails, Sensitive Review Holds and Live Response Preview to its Reviews AI Agents. They aim to give businesses more control before AI replies are published.
Why it matters: If you let AI answer reviews, turn on the guardrails and holds first so a bad reply never goes out under your name.
HighLevel · 26 Oct
HighLevel lets agencies bulk manage Reviews AI (opens the source in a new tab)
Agencies can now manage Reviews AI from Agency, AI Suite, AI Setup. They can set it to Off, Suggestive or Auto, and change one or many sub-accounts at once.
Why it matters: If you run many client accounts, you can now change review reply modes in one place instead of account by account.
HighLevel · 26 Oct
PaperPilot: a Rust PDF engine with WASM and MCP (opens the source in a new tab)
PaperPilot is a PDF engine written in Rust. It runs without the cloud and works with WASM and MCP.
Why it matters: If you handle client PDFs and worry about privacy, a tool that avoids the cloud is worth a look.
github.com via Hacker News
Tools and products
HighLevel Workflows can now trigger on Social Planner posts (opens the source in a new tab)
A new Social Planner Posts trigger in HighLevel Workflows starts automations from post statuses and publishing events. Examples include failed posts and approvals.
Why it matters: You can build alerts for failed posts and approval steps, which cuts manual checking for your team.
HighLevel · 25 Oct
Social comments now appear in HighLevel Conversations (opens the source in a new tab)
Comments on posts published through Social Planner now show in the Conversations Team Inbox next to SMS, email and DMs. Teams can reply, move to private chat and automate follow-ups.
Why it matters: Your team can handle social comments from the same inbox as other messages, so fewer leads get missed.
HighLevel · 26 Oct
HighLevel launches Social Intelligence in limited beta (opens the source in a new tab)
Social Intelligence lets users explore social conversations, monitor topics, watch competitors and get alerts inside HighLevel. It is open only to selected users.
Why it matters: If you get access, you could track your brand and competitors without another paid tool, but it is not open to everyone yet.
Also worth knowing
Apollo economist says AI paid off little outside tech (opens the source in a new tab)
An Apollo economist says AI has delivered zero financial payoff to corporate America outside of tech, despite large investment and hype.
Why it matters: Before you buy more AI tools, set a clear measure of time or money saved for each one.
TechCrunch · 25 Oct
Argus: free open-source agentic security operations platform (opens the source in a new tab)
Argus is a free, open-source security operations platform that uses AI agents. It is released under the MIT licence.
Why it matters: Small teams with no security staff may want to test it, but treat it as unproven until others review it.
thomasunise.com via Hacker News · 24 Oct
Old CNN beats frontier LLMs at skin cancer diagnosis (opens the source in a new tab)
A lab post says a 10-year-old CNN did better than frontier LLMs at diagnosing skin cancer.
Why it matters: The newest general model is not always the best one, so test a specialised tool against it on your own task.
lynchlab.net via Hacker News · 25 Oct