PrismML compresses 27B reasoning model to iPhone scale
PrismML's compression breakthrough enables a 27-billion-parameter reasoning model to run on iPhone hardware while retaining 90 percent of performance, with minimal degradation in math and coding tasks. This directly addresses the on-device AI bottleneck that has constrained Apple and other device makers from deploying capable reasoning locally. Apple's reported testing signals serious intent to shift inference workloads off-cloud, reshaping the competitive dynamics between cloud-dependent and edge-first AI stacks. The achievement matters because it collapses the traditional tradeoff between model scale and deployment footprint, opening reasoning capabilities to billions of consumer devices.
Modelwire context
Analyst takeThe 90 percent performance retention figure is doing a lot of work here, and the benchmark composition matters enormously: a model that holds up on math and coding but degrades on multi-step reasoning or instruction following would still represent a meaningful capability gap from its cloud counterpart, even if the headline number looks clean.
The edge inference story sits in direct tension with where OpenAI is placing its bets. As covered here on July 15, OpenAI's hardware move was a developer peripheral for Codex, a cloud-tethered product that assumes the compute stays remote. PrismML's approach points the other direction entirely: if capable reasoning can run locally, the value of maintaining a cloud inference relationship with end users weakens, and OpenAI's peripheral strategy looks like it is optimizing for a model of distribution that may have a shorter runway than assumed.
Watch whether Apple ships a developer API for on-device reasoning in its next SDK release cycle. If it does, and if PrismML's model is among the supported options, that confirms Apple is treating this as infrastructure rather than a research curiosity.
This analysis is generated by Modelwire’s editorial layer from our archive and the summary above. It is not a substitute for the original reporting. How we write it.
MentionsPrismML · Bonsai 27B · Apple · The Decoder
Modelwire Editorial
This synthesis and analysis was prepared by the Modelwire editorial team. We use advanced language models to read, ground, and connect the day’s most significant AI developments, providing original strategic context that helps practitioners and leaders stay ahead of the frontier.
Modelwire summarizes, we don’t republish. The Decoder originally reported this story as “Bonsai 27B is a full open reasoning model that fits on an iPhone”. The full content lives on the-decoder.com. If you’re a publisher and want a different summarization policy for your work, see our takedown page.