Google's Gemini has deep thoughts; Chainguard shows its work

Today on Product Saturday: Google jumps on the reasoning model train, Chainguard promises to leave no vulnerability unseen, and the quote of the week.

Google's Gemini has deep thoughts; Chainguard shows its work
(Credit: Google)

Welcome to Runtime! Today on Product Saturday: Google jumps on the reasoning model train, Chainguard promises to leave no vulnerability unseen, and the quote of the week.

(Was this email forwarded to you? Sign up here to get Runtime each week.)


Ship it

Flashdance: As chatter about DeepSeek's efficient yet competitive models continues to rattle throughout the market, Google took another step toward carving out a place at the high end. This week it released its Gemini 2.0 Flash model to all developers through Google AI Studio and Vertex AI.

Google described Gemini 2.0 Flash as "a powerful workhorse model, optimal for high-volume, high-frequency tasks at scale and highly capable of multimodal reasoning across vast amounts of information with a context window of 1 million tokens." This year will be an interesting test of how many enterprises feel the need to invest in top-tier models when cheaper ones that get the job done are coming out practically every week.

Into the deep: Speaking of companies that could be vulnerable to the proliferation of cheaper "good enough" large-language models, OpenAI introduced Deep Research this week. The tool was designed to scour the web for information: "give it a prompt, and ChatGPT will find, analyze, and synthesize hundreds of online sources to create a comprehensive report at the level of a research analyst," OpenAI said.

"On the whole I’m impressed with OpenAI’s new tool—at the very least it gives you a framework and some sources and ideas to start you off on your own research," wrote Mark Sullivan at Fast Company. About 24 hours after Deep Research dropped, Hugging Face released Open Deep Research, which "seeks to match Deep Research's performance while making the technology freely available to developers," according to Ars Technica.

Decision support: Chainguard has raised $256 million over the last several years to tackle the software supply-chain security problem by providing trusted images for developing software. This week it released a new feature in Chainguard Images that will help customers prove that approach is working when it comes time for their bosses to approve renewing the subscription.

CVE Visualizations now comes standard in the Chainguard Console and allows customers to "better understand and communicate the benefits of adopting Chainguard Images and the true cost of ongoing CVE management," the company said in a blog post, referring to the "common vulnerabilities and exposures" database. Score one for the marketing department.

The ACID test: The ACID (atomicity, consistency, isolation, and durability) standard is used to describe a database that can handle the unreliability of the internet in order to guarantee the reliability of the data it contains. This week Aerospike introduced the newest edition of its flagship database with support for distributed ACID transactions, which makes the NoSQL database more appealing for revenue-generating applications.

Aerospike Database 8 is "the first real-time distributed database to guarantee strict serializability of ACID transactions at a small fraction of the cost of competing systems," according to Silicon Angle. ACID compliance is a vital requirement for a lot of companies operating in regulated industries (for now, anyway), and should open up Aerospike's database to a wider set of potential customers.

Next please: As noted in our look at Microsoft's widely used Visual Studio Code editor last month, generative AI technologies are changing what software developers want and need from one of their most important tools. This week Microsoft released an update for Visual Studio Code that contains a preview of a feature called Copilot Next Edit Suggestions, which "based on the edits you're making, Copilot NES both predicts the location of the next edit you'll want to make and what that edit should be," the company said.

"GitHub Copilot code completions are great at autocomplete, but since most coding activity is editing existing code, it's a natural evolution of completions to also help with edits," Microsoft said. The new feature lends further credence to the idea that AI coding assistants won't replace software engineers, but could have enormous potential to alleviate a lot of the busywork involved in developing and maintaining software.


Stat of the week

Companies are starting to get the message when it comes to ransomware: Don't feed the trolls. "The total volume of ransom payments decreased year-over-year by approximately 35%," according to a new report from Chainalysis, citing "increased law enforcement actions, improved international collaboration, and a growing refusal by victims to pay."


Quote of the week

“These are highly trained professionals. This is not some roving band running around doing things." — Treasury Secretary Scott Bessent, describing the roving band of racists running around doing things inside government computer systems under the direction of Elon Musk.


The Runtime roundup

Cloudflare's R2 storage service was down for about an hour Thursday thanks to "human error and insufficient validation safeguards during a routine abuse remediation for a report about a phishing site hosted on R2," although no data was lost, the company said in a post-mortem Friday.

If you needed to read 2,000 words describing just how weird that Salesforce Agentforce commercial starring Matthew McConaughey and Woody Harrelson is, Defector has you covered.


Thanks for reading — see you Tuesday!

Great! You’ve successfully signed up.

Welcome back! You've successfully signed in.

You've successfully subscribed to Runtime.

Success! Check your email for magic link to sign-in.

Success! Your billing info has been updated.

Your billing was not updated.