"Part of the reason that we felt the need to do an operating system was because there's a lot of fragmentation in the marketplace," Krisztian Flautner, the firm's vice-president of research and development, told the BBC.
"An important aspect of that is productivity. Instead of having large teams spending years designing a product, we'd like to turn that into months, so that you can take the [hardware] components, assemble the right ones, connect the device and focus on the problem you are solving and not the means to getting there."
While PC and smartphone operating systems typically take up gigabytes of storage, Mr Flautner said mbed OS had been deliberately designed to require a fraction of the amount.
"We're talking a few hundred kilobytes here," he said.
"It's much smaller, but that doesn't mean it's less complex. There's still a lot of software components that have to work together.
"[And] of course when you do it on these very constrained devices, an operating system becomes much more configurable. So, you can choose to leave bits out to reduce the memory footprint."
One expert said he expected that most hardware developers would welcome the release, but cautioned that the OS would not address everyone's needs.
"As critical as the operating system is for a lot of these devices, it is itself no longer a differentiating technology. It's an enabling one," said Chris Rommel from the VDC Research Group.
"It's a platform that is needed to serve as a foundation for internet-of-things functionality and services.
"By offering a solution down to the smallest class of devices that would use M-class chips, ARM's potentially setting up the success of its partners as they pursue IoT technologies.
"However, there will likely never be any one operating system - or even two or three - that can satisfy the broad ranges of needs of all the various devices that compose the internet of things. They are just too different."
ARM said its launch partners for mbed OS included IBM, Telefonica and chipmakers Marvell, NXP and Freescale.