I'd be surprised if the 3.5" MAF was ever over-run by a saab. There are people pulling ~60lb/min through them on DSMs.
As stated, there needs to be scaling done, and on DSM's there is extrapolation that's done based on VE and RPM that allows the guys to run stock 2g mafs past their limits, and same with 3". You wouldn't REALLY want to do this without some dyno/wideband time, but it's doable, and repeatable. I am, by no means, a code-writer, but the concept is easy.
Anyways.. back to the OP. 400hp is not anywhere near the limit of the MAF.
maffrequencylimits [ECMTuning - wiki]
The two most common GM MAF sensors used on our DSMs are the 3" and 3.5", part numbers 25180303 and 25179711, respectively. These sensors operate fundamentally different than the Mitsubishi MAF sensors. The GM MAF sensors work by measuring mass airflow directly. They do not need temperature and baro data because they're not measuring volume. They are reporting air mass directly as a frequency.
The biggest problem, IMO, with these MAF sensors is that this frequency to mass mapping is non-linear. So as airflow increases, it takes less and less change in frequency to present a larger and larger change in air mass. This makes the system very sensitive to fluctuations in frequency output from the sensor.
Neither a MAF Translator nor the ECMLink V3 GM MAF cable allow you to see the raw GM MAF sensor frequency directly. The MAF Translator provides a translation to stock DSM-like frequencies, while the ECMLink V3 GM MAF cable provides a simpler translation that produces raw frequencies that are quite different (but that the ECMLink V3 ECU code knows how to handle). So it's best to avoid talking about limiting-frequencies when discussing a GM MAF sensor when used on a DSM and rather just talk about air-mass metering capability.
Also, the GM MAF sensors, like any MAF sensor, are sensitive to anything that causes the airflow through the sensor to be unevenly distributed across the entire area of the sensor opening. It's very common to see smaller diameter piping leading to or from a GM MAF sensor cause the sensor to produce a higher airflow reading than that actual mass of air flowing through the sensor. Because the metering capability of the sensor is dictated by the maximum airflow reading it can report, having it report a value that is artificially high reduces the metering capability of the system.
That said, you can expect a 3" GM MAF to easily measure (report) 600 gm/sec (80 lb/min) well before any "limit" is reached. The 3.5" would probably measure over 700 gm/sec (92 lb/min)! The problem is that we have very little data to back any of this up. But the bigger problem, again, is that well before you reach these limits, the frequency to airflow mapping will simply become too sensitive to be usable. The result is a very choppy airflow signal to the ECU and many of the same problems mentioned above with MAF overrun on a Mitsubishi MAF. It's just for a different reason.
So with that in mind, we typically recommend that ECMLink users enable the MAF Clamp function with running a GM MAF. Enable it around, say, a MAFRaw reading of 2700-3000hz (that might vary a bit depending on whether you are using a MAF Translator or an ECMLink V3 GM MAF cable) so that the MAF signal is effectively ignored above that frequency anyway and it won't matter what's going on with the signal. This also effectively removes any metering limit that might have existed.
