OpenRouter provider table for meta-llama/llama-3.3-70b-instruct (per-provider price, P50 latency, throughput, uptime)

Fetched 2026-10-14. Per-provider rows for Llama 3.3 70B: DeepInfra (Turbo) $0.100/$0.320 at 0.51s P50 latency / 14 tps; NovitaAI $0.135/$0.400, 0.58s, 41 tps; AkashML $0.200/$0.520, 0.31s, 45 tps; Parasail $0.220/$0.500, 0.44s, 49 tps; Cloudflare $0.293/$2.253, 0.43s, 35 tps; SambaNova Turbo $0.60/$1.20 (25% off -> $0.45/$0.90), 0.77s, 98 tps; Groq $0.59/$0.79 + $0.295 cache read, 0.40s, 233 tps, 100% uptime; CoreWeave $0.71/$0.71, 0.13s, 85 tps; Google Vertex $0.72/$0.72, 4.44s, 98 tps; Together $1.04/$1.04, 0.74s, 43 tps. These are OpenRouter-measured P50 numbers over its own traffic, not vendor claims.