Early engineering samples of the NVIDIA GeForce GTX 2070 have been revealed, with a lower core count than the original model. The GeForce "GTX" label has been used from the GeForce 10 to the GeForce 16 Series GPUs. All products after these GPUs are labeled as GeForce "RTX" series, but it looks like the NVIDIA company has considered putting the GTX 20 series label on RTX 20 series cards internally, as such a model does appear online.
The images released by @unnatural__log show a fancy NVIDIA GeForce GTX 2070 GPU, which is said to be an engineering sample, which also explains why it doesn't use the "RTX" tag.
In terms of design, the NVIDIA GeForce GTX 2070 engineering sample uses the same design as the GeForce RTX 2070. It has the same dual-axis fan cooler and dual-slot design, but the label on the side says "GeForce GTX" and doesn't have the words "GTX 2070" or "RTX 2070" on the front. There are a few missing labels on the graphics card, but other than that, there is no big difference. The display port includes the same HDMI, 2x DP, USB Type-C VirtualLink port, and DVI output. The graphics card should have a standard 8-pin connector, as this was before NVIDIA switched to a 12-pin or 16-pin connector.
The PCB for the NVIDIA GeForce GTX 2070 engineering sample also matches the reference design, but its specs are the biggest change. The RTX 2070 has 2304 cores, while the GTX 2070 ES GPU has 2176 cores and is based on the TU106-400A-A1 GPU with a device ID of "10DE-1F07 - 10DE-12AD". This is the same core configuration used by the RTX 2060 SUPER with the TU106-410-A1 GPU.
Based on user testing, the NVIDIA GeForce GTX 2070 Engineering Sample GPU is flashed with the RTX 2070's BIOS "400A BIOS", which doesn't change the device ID or unlock additional cores, but does provide some extra power to the GPU cores that can be manually adjusted. By overclocking, the GPU can eventually achieve 95% of the performance of the RTX 2070, or 16% faster than the default configuration. This is a pretty good boost considering that modern GPUs are already running at clock speeds that far exceed their listed "peak boost" clock speeds.