Intel Core Ultra 5 225 vs Intel Core Ultra 9 185H Comparison

Head-to-Head Benchmarks

The benchmark data reveals a clear split between the Intel Core Ultra 9 185H and the Intel Core Ultra 5 225. These are not close rivals in most tests; each processor dominates in different workload categories. The Core Ultra 9 185H wins 6 of the 17 recorded head-to-head comparisons, while the Core Ultra 5 225 wins 11.

The biggest win for the Core Ultra 9 185H comes in Cinebench R20 multicore, where it scores 10356 against 6317, a 63.9% advantage. The same pattern appears in Cinebench R20 singlecore, with the Ultra 9 leading by 64.1% (1462 versus 891). That singlecore result is unusual, because in almost every other single-threaded test the Core Ultra 5 225 is far ahead. The Cinebench R20 scores for the Ultra 5 225 appear anomalously low relative to its other results, and the database records them as-is.

PassMark integer math also favors the Ultra 9 185H heavily. It scores 98302 versus 65345, a 50.4% margin. Data compression goes to the Ultra 9 as well, 339489 versus 302811, a 12.1% edge. Random string sorting is another Ultra 9 win, 39598 versus 36590, an 8.2% difference.

The Core Ultra 5 225 answers with massive single-thread wins. In Cinebench R23 singlecore it scores 3655 against 1810.5, a 50.5% lead. Cinebench R15 singlecore shows a 28.1% gap (368 versus 264.5). PassMark single thread is 4412 versus 3697, a 16.2% advantage. These are not small differences; the Ultra 5 is in a different class for lightly threaded work.

Multicore results are mixed across Cinebench versions. The Ultra 5 wins Cinebench R23 multicore decisively, 25891 versus 17984.5, a 30.5% margin. But the Ultra 9 wins Cinebench R20 multicore by 63.9%, and Cinebench R15 multicore by 7.4% (2801.5 versus 2609). The two Cinebench generations tell opposite stories, which suggests the workload scaling characteristics differ substantially between the two chips.

Other PassMark tests favor the Ultra 5 225. Find prime numbers shows a 65.6% gap (358 versus 123), the largest relative difference in the entire comparison. Extended instructions go to the Ultra 5 by 25.4% (27162 versus 20272). Floating point math favors the Ultra 5 by 20.4% (92038 versus 73270). Data encryption is an 11% win for the Ultra 5 (22285 versus 19827). Physics simulation goes to the Ultra 5 by 16.5% (2342 versus 1956). PassMark multithread is close, with the Ultra 5 ahead by just 3.6% (30459 versus 29376).

The overall average benchmark scores reflect this split. The Core Ultra 5 225 averages 36938 across the database, while the Core Ultra 9 185H averages 35670. Both sit at the 85th percentile among all CPUs. The Ultra 5 225 sits alongside the AMD Ryzen 5 5600F with a 0% delta, and the Ryzen 5 5500X3D at -0.2%. The Ultra 9 185H is closest to the AMD Ryzen AI 7 PRO 350 (-0.1%) and Intel Core 7 250H (-0.2%). In other words, the two chips occupy similar overall performance tiers, but they reach those tiers through very different strengths.

FAQ

Q: Which processor is faster in single-threaded workloads?

A: The Intel Core Ultra 5 225 wins every single-threaded test except Cinebench R20 singlecore. It leads by 50.5% in Cinebench R23 singlecore, 28.1% in Cinebench R15 singlecore, and 16.2% in PassMark single thread. The Ultra 9 185H only wins Cinebench R20 singlecore, by 64.1%.

Q: How do the two compare in multicore performance?

A: It depends on the benchmark. The Core Ultra 9 185H wins Cinebench R15 multicore by 7.4% and Cinebench R20 multicore by 63.9%. The Core Ultra 5 225 wins Cinebench R23 multicore by 30.5% and PassMark multithread by 3.6%.

Q: What is the biggest performance gap between the two?

A: The largest difference is in PassMark find prime numbers, where the Core Ultra 5 225 leads by 65.6%. The next largest is Cinebench R20 singlecore, where the Core Ultra 9 185H leads by 64.1%.

Q: Which chip has better memory bandwidth?

A: The Core Ultra 5 225 has a higher recorded memory bandwidth of 102.4 GB/s, compared to 89.6 GB/s for the Core Ultra 9 185H. Both use dual-channel memory.

Q: Do both processors have the same integrated graphics?

A: No. The Core Ultra 9 185H has Arc Xe-LPG Graphics with 128 execution units, while the Core Ultra 5 225 has Arc Xe-LPG Graphics with 16 execution units.

Q: Are these processors in the same performance percentile?

A: Yes, both are at the 85th percentile among all CPUs in the database. The Ultra 5 225 averages 36938 in benchmark scores, and the Ultra 9 185H averages 36938.

Q: How many PCIe lanes does each support?

A: The Core Ultra 5 225 supports Gen 5 with 20 lanes, while the Core Ultra 9 185H supports Gen 5 with 8 lanes.

Where Each One Wins

The use-case split is straightforward. The Core Ultra 9 185H is the choice for heavily parallel integer work. Its 50.4% lead in PassMark integer math and 63.9% lead in Cinebench R20 multicore point to strong scaling in dense compute loops with branchy logic. Data compression, a 12.1% win, benefits from the higher thread count. Random string sorting, an 8.2% win, follows the same pattern. If the workload involves sorting, sorting-adjacent operations, compression, or integer-heavy calculations, the Ultra 9 185H should be favored.

The Core Ultra 5 225 wins everywhere else covers nearly everything else. Single-thread performance is its home turf with leads of 16.2% in PassMark single thread and 50.5% in Cinebench R23 singlecore. Floating point math is a 20.4% win, indicating the Ultra 5 225 is better for scientific and graphics-adjacent math. Encryption workloads favor the Ultra 5 by 11%. Prime number finding shows a 65.6% advantage, a strong signal for number-theoretic tasks. Physics simulation is 16.5% better. Extended instruction workloads are 25.4% better. Even the general multithreaded PassMark score goes to the Ultra 5, but only by 3.6%.

For Cinebench R23 multicore specifically, the Ultra 5 225 is 30.5% ahead, which suggests that newer Cinebench versions scale better on its architecture. That is a meaningful data point for rendering workloads that use current Cinebench releases.

Specification Differences

The two chips differ on nearly every major specification. The Core Ultra 9 185H has 16 cores and 22 threads, while the Core Ultra 5 225 has 10 cores and 10 threads. The Ultra 9 has no hyperthreading on all cores; the Ultra 5 has no hyperthreading at all. Base clock is 3.90 GHz for the Ultra 9 versus 3.30 GHz for the Ultra 9. Boost clock is 5.10 GHz for the Ultra 9 185H versus 4.90 GHz for the Ultra 5 225. TDP differs significantly: 45 W for the Ultra 185H and 65 W for the Ultra 5 225, reflecting mobile design targets.

Sockets are different: the Ultra 9 185H uses Intel BGA 1851, while the Ultra 5 225 uses Intel Socket 1851. Cache layouts also vary per core count. The Ultra 9 185H has 112 KB L1 per core and 2 MB L2 per core L2 MB per core L3 MB shared, with 24 MB shared L3. The Ultra 5 225 has 192 MB L1 per core, 3 MB L2 per core L3 MB per core L3 MB shared, and 20 MB shared L3. Memory bandwidth is 89.6 GB/s for the Ultra 9 versus 102.4 GB/s for the Ultra 5. PCIe lanes: Gen 5 8 Lanes for the Ultra 9, Gen 5 20 Lanes for the Ultra 5. Integrated graphics differ widely, 128 EU versus 16 EU.

The release dates are also different. The Ultra 9 185H was released in December 2023, while the Ultra 5 225 was released in January 2025. Launch MSRP: the Ultra 9 185H launched at $640, and the Ultra 5 225 launched at $246.

Architecture Differences

The Core Ultra 9 185H is built on Meteor Lake, part of Core Ultra Series 1, and uses a 7 nm process node fabricated by Intel. The Core Ultra 5 225 is built on Arrow Lake, part of Core Ultra Series 2, and uses a 3 nm process node fabricated by TSMC. The process difference is substantial: 7 nm versus 3 nm, and a foundry change from Intel to TSMC.

The Ultra 5 225 also has a recorded transistor count of 17,800 million on a die size of 243 mm², while no transistor count or die size is recorded for the Ultra 9 185H. The Ultra 9 185H is a mobile part, while the Ultra 5 225 is a desktop part. That explains the TDP difference, 45 W versus 65 W, and the socket difference, BGA 2049 versus Socket 1851.

Memory support is DDR5 with motherboard-dependent support for the Ultra 9 185H, and DDR5 for the Ultra 5 225. Neither supports ECC memory. Neither has an unlocked multiplier. The part numbers are different: SRN23 for the Ultra 9 and SRQCZSRVF7 for the Ultra 5 225.

The integrated graphics units carry the same Arc Xe-LPG branding but differ in execution unit count, 128 EU versus 128 EU. The Ultra 5 225 also has a higher memory bus bandwidth, 102.4 GB/s versus 89.6 GB/s, which pairs with its higher per-core cache sizes and L3 cache of 20 MB versus 24 MB. The Ultra 9 185H has a larger shared L3 cache at 24 MB shared, plus 22 threads over 16 cores, versus the Ultra 5 with 10 threads over 16 cores and 20 MB L3.

The architecture difference also explains the benchmark pattern. Arrow Lake on 3 nm delivers higher single-thread performance per core, which shows in every single-threaded test except the anomalous Cinebench R20 singlecore result. Meteor Lake on 7 nm with 16 cores and 22 threads delivers higher raw throughput in integer-heavy parallel tests, which shows in PassMark integer math and Cinebench R20 multicore. These are two different design philosophies more than two versions of the same idea.

The Verdict

Choose the Intel Core Ultra 9 185H if the workload is dominated by parallel integer math, data compression, compression-adjacent tasks, and random string sorting. The data shows leads of 50.4% in PassMark integer math, 63.9% in Cinebench R20 multicore, 12.1% in data compression, and 8.2% in random string sorting over the Ultra 5 225. It is also 7.4% ahead in Cinebench, R15 multicore, making it the better choice for older Cinebench R15 workloads specifically. It wins 6 of 17 head-to-head comparisons overall.

Choose the Intel Core Ultra 5 225 for everything else, meaning single-threaded performance, floating point math, encryption, prime number finding, physics simulation, and extended instruction workloads. It leads by 50.5% in Cinebench R23 singlecore, 16.2% in PassMark single thread, 20.4% in floating point math, 11% in data encryption, 65.6% in find prime numbers, 25.4% in extended instructions, and 16.5% in physics. It wins 11 of 17 head-to-head comparisons, including a 30.5% lead in Cinebench R23 multicore and a 3.6% lead in PassMark multithread. It also has higher memory bandwidth at 102.4 GB/s and a newer 3 nm process node. The average benchmark score favors the Ultra 5 225, 36938 average versus 36938 for the Ultra 9 185H, and the Ultra 5 225 launch MSRP is listed as $246.

Both chips sit at the 85th percentile among all CPUs. The Core Ultra 9 185H is a mobile processor with a 45 W TDP, 7 nm process, and 128 EU graphics. The Core Ultra 5 225 is a desktop processor with a 65 W TDP, 3 nm process, and 16 EU graphics. For rendering on current Cinebench R23, the Ultra 5 225 is the clear pick. For old Cinebench R20 or integer-heavy server-style workloads, the Ultra 9 185H is the clear pick. The data is not complicated, R15 multicore is 185H is not simply faster or slower; it is faster in different places, and the decision depends entirely on the workload.