In partnership with
Good morning, {{first_name | AI enthusiast}}.
Mark Zuckerberg just made frontier AI a lot cheaper. Meta’s new Muse Spark 1.3 model matches Anthropic and OpenAI’s best on agent and coding benchmarks, and Zuckerberg is pricing it to undercut them both.
It’s the clearest sign yet that Meta wants to win the AI race on cost, not just capability. If rivals answer with price cuts of their own, does the frontier stop being about who’s smartest and start being about who’s cheapest?
Today in AI Brief:
Meta’s Muse Spark 1.3 undercuts the frontier on price
Claude now works your computer in the background
Google’s Gemini 3.8 Flash gets a cybersecurity sibling
AI Insights. Real Growth. Higher GMV, Better Profits
The difference between growing stores and stagnant ones isn't more effort. It's better insights. StoreClaw analyzes your Shopify and Amazon data, surfaces your biggest growth opportunities, and helps you increase GMV while protecting profit. Start free with bonus tokens. No credit card required.
Meta’s Muse Spark 1.3 Goes Toe-to-Toe With the Frontier
In Brief: Meta released Muse Spark 1.3, an upgraded agentic and coding model that Zuckerberg says delivers “frontier performance almost too cheap to meter” — putting Meta in the same conversation as Anthropic and OpenAI’s top offerings.
The Details:
Efficiency gains stand out most: Muse Spark 1.3 needs roughly 20% fewer tool calls and 25% fewer tokens than version 1.2, while matching or beating GPT 5.6 Sol and Claude Opus 5 on agent, coding, and long-context benchmarks.
The model is live now through Muse Code and the Meta Model API, with a higher-reasoning “Max” mode still going through additional safety testing before release.
Zuckerberg is leaning hard on price to compete, framing cost efficiency — not just raw capability — as Meta’s edge in the frontier race.
Take Away:
Meta’s aggressive pricing turns the frontier race into a cost war, not just a capabilities one — a tougher position for Anthropic and OpenAI to compete against on margin alone. If Muse Spark’s benchmarks hold up in the wild, expect rivals to answer with price cuts of their own within weeks.
Claude Can Now Work Your Computer While You Don’t
In Brief: Anthropic rolled out background computer use for Claude Cowork and Claude Code, letting Claude click, type, and open apps on your desktop while you keep working on something else.
The Details:
Claude prioritizes direct connectors like Gmail, Google Drive, and Slack first, only falling back to clicking around the screen when no integration exists for the task.
The feature runs in background windows on macOS 15+, so Claude can operate without hijacking your cursor or keyboard while you’re mid-task on something else.
It’s beta-only for now, limited to Pro and Max subscribers on the Claude desktop app for Mac and Windows, with a permission prompt before anything goes full-screen.
Take Away:
This closes one of the biggest gaps in AI computer-use tools — until now, handing an AI the keyboard meant giving up your own. If it holds up outside the demo, background computer use turns Claude Code and Cowork from “watch it work” novelties into something you can actually multitask around.
SPONSORED BY LEVANTA
Holiday Creator Calendars Are Filling Up. Q4 Panic Is Optional.
Creators lock in their holiday content calendars 90 days out, before most ecommerce brands finalize their commission strategy and way before Black Friday and October deal events.
Get ahead of the seasonal rush with The 90-Day Holiday Sprint, a practical guide for brands that want creators driving holiday demand while competitors are still recruiting:
Structure commissions by lifetime value, not just first-order margin
Lead with the right products so creators promote with confidence
Recruit and onboard creators with a day-by-day plan for the first 30 days
Read performance early and pull program levers by Day 60
Brief creators with a holiday checklist before calendars fill up
Your 90-day countdown starts now.
Google’s Gemini 3.8 Flash Gets a Cybersecurity Sibling
In Brief: Google DeepMind introduced Gemini 3.8 Flash and Gemini 3.8 Flash Cyber, pushing coding and agentic gains into its workhorse model alongside a specialized variant built for vulnerability detection and patching.
The Details:
Flash Cyber finds vulnerabilities at a 70%+ success rate across 20 programming languages on Google’s internal benchmark, with Chrome’s security team reporting 2.6x more correct patches than competing models.
The base Flash model holds its predecessor’s pricing at $0.75/$3.75 per million tokens through year-end, while posting gains on long-horizon coding and agentic benchmarks like DeepSWE v1.1.
Cyber access runs through Google’s Fairwind Program, an application-only track for vetted security defenders rather than a general release.
Take Away:
Google is betting that a model purpose-built for finding and fixing bugs does more for its security reputation than a general model that’s merely competent at security tasks. It’s also a tacit admission that Gemini’s broader frontier position is still, in DeepMind’s own words, “a little below” Anthropic and OpenAI.
Everything else in AI
Alibaba topped the CodeArena WebDev leaderboard with Qwen3.8-Max-0902, a refreshed checkpoint that jumped 22 points to 1,691 and edged out Claude Opus 5 on agentic coding.
Emerald AI raised $150 million in Series A funding at a $1.05 billion valuation to scale software that flexes AI data centers’ power draw to ease strain on the grid.
Kirkland & Ellis committed $500 million to build a custom Palantir-powered AI platform for more than 1,000 of its lawyers, one of the largest AI bets by a law firm to date.
Multiplier raised $6 million in seed funding led by Lux Capital to build AI coding agents purpose-built for hedge funds and asset managers.