Viral AI News (Sep 30): Sol's Price Cut, Opus Nerf & No MCP

GPT-6.1 Sol hit 1,017 Hacker News points, a Dots clone took 1,569 stars in a day, and Livenerf asks if Opus 5.5 was nerfed. The Sept 30 viral AI news, verified.

Dark editorial illustration of a viral AI news post spreading across social media feeds

OpenAI's DevDay produced the biggest thread on Hacker News, a clone repo that found 1,500 GitHub stars in a single day, and a pricing line that developers cared about more than the model itself. Meanwhile the loudest argument in the Claude community is still whether the model got quietly worse.

Here is what actually went viral in AI over the last 24 hours, and how much of it holds up.

The Sol thread owned Hacker News, and the price did the work

The submission announcing GPT-6.1 Sol hit 1,017 points and 890 comments on the Hacker News front page, by far the highest-scoring AI story in this window. The model is OpenAI's cheaper tier: $2 per million input tokens, $10 per million output and $0.10 for cached input, which OpenAI says delivers near GPT-6 Astra quality for agentic coding, computer use and professional work.

That cached input price is the part engineers fixated on, since it decides what a long agent run actually costs. OpenAI shipped Sol one week after GPT-6 Sol and one day after scrapping the planned GPT-6.1 Astra over safety concerns. Source: OpenAI Source: CNBC Hacker News thread

Livenerf: the nerf question, now with a pre-registered benchmark

"Has Opus 5.5 been nerfed yet?" reached 780 points and 329 comments. Livenerf is an independent, unaffiliated test harness that runs a locked panel of 78 questions against Claude Opus 5.5 every day, with an Opus 5 and GPQA control arm, and publishes raw logs plus paired per-item statistics over 10-day windows. The author says the panel was calibrated from 2,336 screened questions and that no single day's result should drive decisions; a deviation above 7.5 points is pre-registered as the threshold for calling a change.

The honest answer is that no nerf has been detected. This is community-sourced methodology, not a finding, and the repo documents day six rather than a proven regression. What it does prove is that users no longer trust silent model updates, and someone is finally measuring them in public. GitHub repo Hacker News thread

Somebody cloned Dots in a day and got 1,569 stars

OpenAI launched Dots, always-on agents that each run their own cloud computer and reach more than 4,000 apps, to 699 points and 566 comments. While that thread argued about agent runaway and privacy, an open-source web clone appeared on GitHub the same day and collected 1,569 stars in under 24 hours.

The clone is a fraction of the real product, and the reviews were split between people calling Dots a genuine shift and people calling it an expensive wrapper. Both reactions are the point: OpenAI shipped a paid, permissioned agent platform and the community immediately tried to build the free version. Source: Reuters GitHub clone Hacker News thread

Pi.dev ate its own words on MCP

Earendil shipped MCP support into the core of its Pi coding agent, a year after marketing a proud no-MCP position, and the post explaining the reversal hit 369 points and 193 comments. The company concedes in writing that MCP is hard to compose, and says it would rather shape the standard from inside than sit it out.

Codemode is a JavaScript sandbox the team added so agents can compose tool calls before they fire. That is the interesting part of the U-turn: a vendor that spent a year arguing against MCP now thinks the fix is a runtime layer on top of it. Source: Earendil Hacker News thread

Quick Comparison

Trend What Went Viral Engagement Verdict
GPT-6.1 Sol Cached input price cut 1,017 points, 890 comments Real, and the price matters most
Livenerf benchmark "Is Opus 5.5 nerfed?" 780 points, 687 GitHub stars Methodology, not proof of a nerf
Dots Open-source clone in a day 1,569 stars, 566 comments Hype with a real product behind it
Pi.dev MCP reversal A vendor abandoning "no MCP" 369 points, 193 comments Honest reversal, worth reading

Frequently Asked Questions

What is the viral AI news today?

OpenAI's DevDay releases dominated every AI thread on Sept 29 and 30. The GPT-6.1 Sol announcement hit 1,017 points on Hacker News, an open-source clone of the new Dots agent collected 1,569 GitHub stars in a day, and a community benchmark asking whether Opus 5.5 got worse reached 780 points.

Has Claude Opus 5.5 actually been nerfed?

No nerf has been detected in public measurements. A community harness called Livenerf runs 78 fixed questions against Opus 5.5 daily and publishes raw logs with paired statistics, using a 7.5 point deviation threshold. Its author warns that single-day results are not decision-grade, and no threshold breach has been reported.

Why are people cloning OpenAI's Dots already?

Dots is the first mainstream paid product for always-on agents. Each Dot runs its own cloud computer, browses the web and connects to thousands of apps, which raises obvious questions about permissions and runaway behaviour. A web clone appeared within hours and collected 1,569 GitHub stars in under a day.

More viral AI coverage | Yesterday's edition: Jev Clones, Tracker Leak & Meta's Muse

If this story matters to you, these earlier editions cover the same ground in more detail:


Building an AI agent, automation, or backend workflow? AsaanKaroobar builds custom AI solutions, apps, and digital experiences. Get in touch →