GroqCloud Llama 3.3 70B Versatile model page (enterprise-gated, vendor speed)

Fetched 2026-10-17. Model id llama-3.3-70b-versatile, badged Enterprise. Vendor token speed about 280 tps (not used as measured_tps; the record keeps the Artificial Analysis median). Context window 131072, max output 32768. No public per-1M price on the page. The 2026-10-14 models index still showed Contact Sales for this model, and this page does not add a public rate. Quantization note: Groq TruePoint Numerics.