Viral

Anthropic's Mythos 5.1, Google Pics vs Canva & Jio's $11 AI PC: Viral AI News for September 2, 2026

Anthropic ships Mythos 5.1, Google launches a Canva killer, Jio turns old PCs into AI machines for $11, and AfterQuery's viral 5-month unicorn run makes headlines.

Daniyal

· 5 min read
Abstract visualization of viral internet trends and AI models with neon pink and electric blue energy bursts

Anthropic quietly shipped the most capable cybersecurity and biology model in existence, then priced it like a premium API. Google launched a Canva killer that lives inside Workspace. Reliance's Jio figured out how to turn any old computer into an AI PC for $11. And the internet is losing its mind over how fast AfterQuery went from YC batch to $3.2 billion. Here is the viral AI news making waves across the industry today.

This Week's Viral AI News: Mythos, Google Pics & JioPC

Catch up on yesterday's Viral AI News before reading today's picks.

Anthropic Ships Mythos 5.1 and Fable 5.1

On September 1, Anthropic released Claude Mythos 5.1, its most capable model for cybersecurity and biology research, alongside Fable 5.1, the same underlying model with safety guardrails for broader availability.

The pricing tells the story: $10 per million input tokens and $50 per million output tokens for Mythos. That is premium pricing for premium capabilities. The model is available only to vetted organizations through trusted access programs, though Anthropic is working to expand access.

The interesting detail is in the safeguards. Fable 5.1 can now identify software vulnerabilities in source code (something Fable 5 could not do), and its biology safeguards intervene on benign requests 85% less often than the original Fable 5 launch. This is Anthropic threading the needle: making the model more useful for legitimate security researchers while keeping dangerous capabilities restricted.

Claude Security now runs on Mythos 5.1.

Sources: Anthropic, Neowin

Google Pics Launches as a Canva Rival Inside Workspace

Google shipped Google Pics to general availability on September 1, a prompt-based image creation and editing tool built directly into Google Workspace. Powered by Google's Nano Banana image-generation model, it opens as an overlay inside Docs and Slides or as a standalone tool at pics.new.

The feature set is real: object-based editing, text-in-image translation, cropping for web/social/print, upscaling to 2K or 4K, and collaborative sessions with reversible batch changes. Google Workspace customers and Google AI Pro or Ultra subscribers get access.

The competitive angle is sharp. Canva has been expanding aggressively into AI and marketing tools, while Google is embedding creative AI directly into the productivity suite millions of businesses already use. No separate app, no extra subscription, just an image editor that appears when you click an image in a document.

Sources: TechCrunch, 9to5Google

Jio Turns Old PCs Into AI Machines for $11

Reliance Jio launched JioPC, a cloud-based PC experience that turns aging computers into AI-ready machines for as little as $11 for two months. The service streams a full desktop environment from the cloud, meaning even a decade-old computer can run modern AI tools through a browser.

This is classic Ambani strategy: take a massive existing market of underpowered hardware and add value through infrastructure rather than new devices. India has hundreds of millions of aging PCs that cannot run modern AI workloads locally, and Jio is betting that cloud delivery makes the hardware irrelevant.

The move comes as Reliance has earmarked 10 trillion rupees (roughly $120 billion) for AI investment, positioning Jio at the center of India's AI ecosystem.

Sources: Yahoo Tech, TechCrunch

AfterQuery's Five-Month Journey to $3.2B Breaks Records

The speed of AfterQuery's valuation climb has the startup world talking. The AI training-data startup went from a $300 million Series A in April to a reported $3.2 billion valuation in September. Five months, more than 10x.

Y Combinator partner Gustaf Alströmer confirmed it is the fastest any startup in the accelerator's history has gone from launch to unicorn status. The founders, now 22 and 23, attended YC's Winter 2025 cohort just 18 months ago.

The company hit a $100 million annualized revenue run rate by April. It employs doctors, lawyers, and other specialists to train AI models on professional reasoning patterns, not just correct answers. Customers include Nvidia and Korean AI lab Motif Technologies.

The valuation raise puts AfterQuery in territory usually reserved for infrastructure and model companies. Whether that holds depends on whether "professional reasoning data" proves to be a durable moat or a commodity that scales with AI itself.

Sources: TechCrunch, Forbes

Anthropic Drops Retailer Agent Blueprints Ahead of Holiday Season

Anthropic announced agent blueprints for retailers on September 2, giving shopping and merchant companies templates to build Claude-powered agents for the holiday shopping season. The blueprints cover two use cases: shopper agents that make preference-based suggestions and add items to carts, and merchant agents that advise on inventory, pricing, and marketing.

The timing is deliberate. Holiday shopping is the make-or-break season for retail, and Anthropic is betting that retailers want to deploy AI agents but need templates rather than starting from scratch. The agents do not complete purchases on behalf of shoppers or merchants, a deliberate restraint.

Sources: Reuters

The pattern this week is clear: the most interesting moves are not the biggest headlines. They are the infrastructure plays, the pricing strategies, and the quiet product launches that reshape how AI gets used in the real world.

Frequently Asked Questions

What benchmark scores did Anthropic's Fable 5.1 and Mythos 5.1 achieve?

Fable 5.1 scored 55.8% on Terminal-Bench 4.0 and 52.6% on Terminal-Bench-Science 0.1, more than doubling Fable 5's 24.7% on the latter. Mythos 5.1 reached 60.9% on Terminal-Bench 4.0 under permissive cyber safeguards. On CursorBench 3.2.0, Fable 5.1 achieved 73.4%, placing it among the top coding agents. These scores represent the highest performance Anthropic has reported on agentic benchmarks.

How much did Fable 5.1's cache read pricing drop and what are the savings?

Fable 5.1 cut cache read pricing by 75%, dropping from $1.00 to $0.25 per million tokens — just 2.5% of the standard input price instead of the usual 10% multiplier. Anthropic estimates typical workloads run about 25% cheaper than Fable 5, while highly agentic tasks save up to 45%. The base rate of $10/$50 per million input/output tokens remains unchanged.

How do Fable 5.1 and Mythos 5.1 safeguards differ?

Fable 5.1 and Mythos 5.1 share the same underlying model but differ in safety controls. Mythos 5.1 runs with relaxed cybersecurity classifiers for vetted organizations only, while Fable 5.1 includes full consumer-facing safeguards. Fable 5.1 now reduces cybersecurity false-positive blocks by 60% and biology safeguards intervene 85% less often on benign requests. Mythos 5.1 is restricted to Anthropic's trusted access programs for cybersecurity and life sciences.

Deeper context from the archive on the companies and technologies in this story:

More on Viral

Browse every Viral story →