NVIDIA GeForce RTX 5060 Ti officially unveiled: Blackwell architecture for the mid-range

With the GeForce RTX 5060 Ti, NVIDIA is bringing its new Blackwell architecture to the particularly price-sensitive mid-range segment. The new models will be available in stores from April 16, 2025 – in two variants with 8 GB and 16 GB GDDR7 memory. The target group is primarily gamers who are planning to upgrade from older generations such as the RTX 3060 or RTX 2060, but do not see the need for high-end GPUs such as the RTX 5080 or RTX 5090. Charm offensive for the mid-range or proactive prevention against a possible Radeon RX 9060?

Technically, the RTX 5060 Ti brings many features of the Blackwell architecture to more affordable regions. These include the new 5th Gen Tensor Cores with FP4 support, the 4th generation of ray tracing units and DLSS-4 technology with multi-frame generation. The new Reflex 2 feature for latency reduction is also integrated. According to NVIDIA, the card thus offers up to twice the frame rates and significantly lower latencies compared to the previous generation, which is particularly noticeable in combination with DLSS 4 and ray tracing.

In the 16 GB version, the RTX 5060 Ti is used with eight GDDR7 modules of 2 GB each, which are distributed on both sides of the PCB using the clamshell method. Despite this configuration, the 128-bit memory connection remains technically identical to the 8 GB version. This does not result in a higher memory bandwidth even with double the capacity, but does provide an advantage for memory-intensive applications that require more than 8 GB of VRAM. The memory bandwidth is 448 GB/s at an effective clock rate of 28 Gbps – a significant advantage over the RTX 4060 Ti, which only achieves 288 GB/s with GDDR6. This is made possible by the introduction of GDDR7 memory, which not only promises higher transfer rates, but also more efficient energy management.

The other features are also impressive: With 4,608 CUDA cores, 144 Tensor units and 36 RT cores, the card is clearly positioned above the RTX 4060 Ti. At the same time, the TDP remains comparatively moderate at 180 watts. PCI Express 5.0 is now being used as the interface for the first time in this segment, which can be particularly advantageous for future applications and in memory-limiting scenarios. The card continues to use a PCIe x8 layout.

In terms of outputs, the reference design offers three DisplayPort 1.4a ports and one HDMI 2.1a port. However, manufacturers such as MSI, ASUS and Gigabyte will also offer factory-adapted designs with slightly higher clock rates and extended cooling solutions. According to the official announcement, the new GPU is priced at 379 US dollars (8 GB) and 429 US dollars (16 GB), which is significantly lower than the corresponding variants of the RTX 4060 Ti, which were priced at 399 and 499 US dollars respectively at launch. NVIDIA is thus specifically targeting users for whom the RTX 4070 is priced too high, but who still want to benefit from modern ray tracing and DLSS 4.

It is interesting to note that, according to NVIDIA’s own presentation, the performance of the RTX 5060 Ti can almost triple that of the RTX 3060 Ti – especially with DLSS 4 enabled. In well-known games such as Alan Wake 2, Cyberpunk 2077 or Hogwarts Legacy, frame rates of over 100 FPS should be possible at full ray tracing load, as the diagrams in the presentation show. But beware: Only the RTX 5060 Ti uses DLSS4 and MFG X4, which of course puts the results into perspective. The RTX 4090, beaten by the RTX 5070, politely sends its regards…

For many users, the 16 GB version of the RTX 5060 Ti should be a particularly welcome alternative. While 8 GB is too limited in many current titles in 2025 – especially in WQHD or with high-resolution texture packages – the larger version offers significantly more leeway. Although the memory bandwidth remains the same, the larger capacity means that swapping to the slow system memory can be avoided, which not only prevents jerks but also stabilizes frame times.

At the same time, the RTX 5060 Ti is clearly positioned below the RTX 5070, which offers a wider memory connection with 192 bits at 12 GB VRAM. Despite a nominally lower capacity, the RTX 5070 is therefore significantly faster in memory bandwidth-critical applications. I will also cover this topic in detail again tomorrow in my detailed launch article on April 16, 2025 at 03.00 pm. In addition to comprehensive benchmarks, the focus there will be on the efficiency, thermal implementation and real-world performance of the memory architecture.

The target group for the RTX 5060 Ti is clearly gamers who are looking to upgrade from older generations such as the RTX 3060 or RTX 2060 but do not want to or cannot make the step into the high-end segment in terms of price or technology. Users who have been using a GTX 1060 or 1660 up to now will also receive a powerful, future-proof solution that fully supports the latest technologies such as DLSS 4, Reflex 2 or ray tracing – and at a price that has rarely been achieved in the mid-range segment with comparable features.

With the RTX 5060 Ti, NVIDIA is not only bringing new technologies to the market, but also responding to criticism of the pricing of the previous generation. Whether the card meets these expectations in practice and what the prices will be in reality remains to be seen in extensive tests and on the market.