Created by Nick Herres somewhere around 2000, 2001.   Rendering of the Descent 3 Phoenix Intercepter done in Bryce3D.  Additional credit goes to Kumen who's 3DsMax model was used as a reference. 

http://www.visualarray.com/artwork/Phoenix.html

Copyright belongs to Nick Herres.

Key differences:

There's a 2-seater planned for the T-50;
The T-50 has a slightly larger (78.8 sq. m. against F-22's 78.04) and better-loaded (330 kg/sq. m. against 340, max. 470 vs. 487) wing, which makes it more agile in dogfight, combined with...
3D thrust-vectoring engines (Saturn Make 117 - no official name for the engines, they're still referred by development code) - the F-22's engines are 2D-vector only;
Improved radar signature, claimed to be smaller than F-22's (0.3-0.4 sq. m. in centimetre-range radar waves vs. F-22's 0.4-0.7);
Higher thrust:weight ratio (traditional for Russian fighters) - 1.19-1.38 combat-loaded, 0.84-0.97 max. weapons+fuel load/max. weapons load (1.17 combat/0.88 max. load for F-22);
Higher internal fuel load, 12500 kg., allowing the T-50 to work as a long-range interceptor with a high supersonic dash speed;
Higher maximum airspeed - 2100-2600 Km/h depending on altitude (most planes can actually go faster than their maximum rated airspeed, but at the risk of airframe damage);
Lighter than the F-22 (yes it's larger, but it also weighs less): 18500 kg. empty (F-22's 19820);
Higher max. G-load (10-11 g against 9.5 for the F-22).

If anything the performance is bound to be improved - the numbers are given for the T-50 prototype, not the production version (compare the T-10 with any Su-27 production variant).

Overall it's pretty similar, but the T-50 is designed to be an agile dogfighter with 3D thrust vectoring, a concept the USAF has forfeit. It has fully mobile fins (whole surfaces move, there're no rudders).

Uhm. Yeah.

96/24 FLAC file