AMD has announced the Radeon AI Pro R9700, a model designed specifically for artificial intelligence applications. This product marks the inaugural release in the new "Radeon AI Pro" series, featuring a substitution of the prefix "R" with "W." The model number is prefixed with an "R," and distribution will be managed by board partners.

The theoretical computing power (Compute FP16) associated with the Navi 48 GPU falls short compared to the leading models of the W7000 series (Navi 31). Additionally, the W7900, equipped with 48 GB, provides 50% more memory. AMD guarantees 1,531 TFLOPS in AI applications utilizing the INT4 data format, closely matching the RX 9070 XT's performance of 1,557 TFLOPS.

The Radeon AI Pro R9700 offers double the memory compared to the Radeon RX 9070 XT, which features the same chip. This enhancement enables the local execution of popular AI models that are not feasible on a 16 GB graphics card. Nonetheless, utilizing 32 GB proves advantageous in comparison to 16 GB, highlighting the absence of a more substantial RDNA 4 GPU. With 32 GB of memory, the performance does not attain the level of the previous W series, which offers a maximum of 48 GB.

The AMD Radeon AI Pro R9700 is anticipated to be released by all major board partners beginning in July; however, AMD did not disclose a price yet.