DeepInfra OpenAI-compatible models list (live per-model pricing and context)

Fetched 2026-10-17, unauthenticated GET of the public models list. meta-llama/Llama-3.3-70B-Instruct-Turbo: input_tokens 0.10, output_tokens 0.32 (USD per 1M), context_length 131072, no cache rate, tag chat. deepseek-ai/DeepSeek-V3.1: input_tokens 0.25, output_tokens 0.95, cache_read present, context_length 163840. That is the same 0.25/0.95 the 2026-10-14 pricing-page transcription recorded, so that representative row did not move. Same payload also lists openai/gpt-oss-120b-Turbo at 0.15/0.60 with context 131072. This endpoint is the machine-readable catalog; the marketing pricing page is the human view of the same numbers.