Apple is in talks with PrismML, a Khosla Ventures-backed startup, to evaluate AI model compression technology that could enable powerful AI models to run directly on iPhones, PrismML CEO Babak Hassibi told CNBC. PrismML publicly released compressed versions of Alibaba's Qwen model on Tuesday, reducing the model from roughly 54 GB to less than 4 GB while maintaining all 27 billion parameters for iPhone 15 or newer devices. The discussions aim to address constraints in Apple's AI strategy as the company opened the public beta of iOS 27 one day earlier, giving iPhone owners access to an overhauled Siri while keeping more processing on-device. Hassibi characterized the talks as "very early" but said "things are progressing nicely," with Apple and other companies evaluating the technology's speed, energy efficiency and performance. The technology could reduce cloud-computing costs and support Apple's privacy positioning by allowing AI features to work without internet connections.
PrismML Releases Compressed Qwen Model with 54 GB to 4 GB Reduction
PrismML, a spinout from the California Institute of Technology, reduced Alibaba's open-source Qwen model from roughly 54 GB to less than 4 GB on Tuesday. The compression allows all 27 billion parameters to run on an iPhone 15 or newer device. The startup shrinks AI models by simplifying how internal information is stored, reducing each value from 16 bits to just one or three possible values. Hassibi compared the approach to the chip industry's move from eight-bit to four-bit computing.
The compressed models use between 10 and 15 times less memory, generate responses six to eight times faster and consume three to six times less energy than conventional versions running on existing hardware, according to PrismML. Hassibi acknowledged the models typically lose a few percentage points of overall performance, with factual recall weakening before skills such as reasoning, math and coding. PrismML is releasing two compressed versions of the model for free, designed to run on iPhones, MacBooks and Nvidia-powered PCs.
Apple Evaluates On-Device AI Processing to Reduce Cloud Dependency
"They're really evaluating our technology right now," Hassibi said of Apple. The discussions remain unclear where they will lead, but Hassibi said things are progressing. Apple can send complex requests to cloud-based models, but running more AI directly on the iPhone would reduce the delay associated with sending data to a remote server, lower cloud-computing costs and support the company's privacy pitch. The approach would also allow certain features to work without an internet connection.
Carolina Milanesi, president and principal analyst at Creative Strategies, said smaller models could let Apple move more demanding features onto the iPhone, including computational photography, video generation and health or fitness tools that rely on sensitive personal data. "The more you can do on device, the better it is," she said, pointing to health and medication data that users would want to keep private. Apple already runs parts of its AI system locally, including translation, some summarization and features tied closely to personal information.
PrismML Raised $16.25 Million Seed Round in March
The technology emerged from Hassibi's research group at Caltech. The university owns the underlying patents and licenses them exclusively to PrismML. In March, the company raised a $16.25 million seed round backed by Khosla Ventures and other investors. Hassibi said Google's open-source Gemma model is next in the pipeline, followed by much larger models, including those from frontier labs that today generally require datacenter hardware.
The technology could ultimately extend well beyond phones and laptops to robotics, autonomous systems and other products that need to make decisions quickly without relying on a cloud connection, according to PrismML. "It's very important that the intelligence be local and that it can run fast," Hassibi said. Horace Dediu, founder of Asymco, said Apple is likely trying to keep the large majority of common Siri interactions on-device while reserving the most demanding tasks for the cloud.
Analysts Question Battery Impact and Real-World Performance
Tarun Pathak, research director at Counterpoint Research, said the model's performance on lengthy prompts, battery consumption during multitasking and reliability across millions of requests will be critical. "The ultimate test will be millions of queries, thousands of device combinations and robust testing at scale," Pathak said. Phil Solis, who leads IDC's research on client processors, said power consumption may be the biggest open question. A model that is capable enough to be used frequently or continuously in the background for agent-like tasks could drain a phone's battery even if it requires less memory.
Gil Luria, an analyst at D.A. Davidson, said shrinking models would not eliminate the need for processors or memory. It could simply move more of those chips from datacenters into phones and other devices. "It's not that you're not going to need the chip," Luria said. "You're still going to need the GPU, and you're still going to need the memory." He added that running AI on individual devices can actually be less efficient than using shared datacenter infrastructure because chips in phones may sit idle much of the time.
Morgan Stanley estimates Apple's average dynamic random access memory cost per bit could rise roughly 190% year over year in fiscal 2027, with NAND costs up about 180%. The firm expects Apple to raise the starting price of comparable iPhone 18 models by about $200 to protect margins. Micron shares plunged in March after Google published its TurboQuant paper on cutting memory use without hurting model performance, though the stock later recovered. Pathak said the combination of cloud and on-device AI can serve a more complete, efficient and privacy-centric AI experience, with complex tasks offloaded to the cloud and sensitive, latency-critical tasks executed on-device.
FAQ
What did PrismML release on Tuesday?
PrismML publicly released compressed versions of Alibaba's open-source Qwen model on Tuesday. The company reduced the model from roughly 54 GB to less than 4 GB, allowing all 27 billion parameters to run on an iPhone 15 or newer device.
How does PrismML's compression technology work?
PrismML shrinks AI models by drastically simplifying how internal information is stored, reducing each value from 16 bits to just one or three possible values. The compressed models use between 10 and 15 times less memory, generate responses six to eight times faster and consume three to six times less energy than conventional versions, according to the company.
Why is Apple evaluating on-device AI processing?
Running more AI directly on the iPhone would reduce the delay associated with sending data to a remote server, lower cloud-computing costs and support Apple's privacy positioning. The approach would also allow certain features to work without an internet connection, according to the source.
