I share neither Anirban's nor Rangarajan's opinions ...
The C6 glioma model in rats is by far too immunogenic and the tumor take never reaches 100%. If you want an agressive model in rat, you can use the 9L gliosarcoma (see the attached article by Lefranc et al., Neurosurgery 2003). In syngeneic mice, you can use the G261 glioma model (see Vershuere et al., Int J Cancer 2015).
With respect to human gliomas, you must use those models that diffusely invade the brain parenchyma, which is never the case with the U87 model. The U87 model is excellent to study anti-angiogenic compounds (Lamour et al., Int J Cancer 2010) but it actually does not mimik what happens in clinics (see Camby et al., JNEN 2002).
The Hs683 oligodendroglioma and the U373 astroglioma models mimik relatively well what happens in clinic, both in terms of spreading in the brain and chemosensitivity (see Branle et al., Cancer 2002; Lefranc et al., Neurosurgery 2008; Mathieu et al., Neoplasia 2008).