InfraRabbit

Inference host prices: sambanova

Open-weight models served by inference hosts, with each host's price per 1M tokens (input, output). One row per model, host and quantization.

Sources for this list

SKUNameInput / 1MOutput / 1MCached inputContext windowLast verifiedSource
deepseek/deepseek-chat-v3.1/sambanova/unknowndeepseek/deepseek-chat-v3.1 via SambaNova USD 3.00USD 4.50–131,072 2026-10-24 06:58 UTCsource
deepseek/deepseek-v3.2/sambanova/unknowndeepseek/deepseek-v3.2 via SambaNova USD 3.00USD 4.50–32,768 2026-10-24 06:58 UTCsource
google/gemma-4-31b-it/sambanova/unknowngoogle/gemma-4-31b-it via SambaNova USD 0.38USD 1.15–131,072 2026-10-24 06:58 UTCsource
meta-llama/llama-3.3-70b-instruct/sambanova/unknownmeta-llama/llama-3.3-70b-instruct via SambaNova USD 0.60USD 1.20–131,072 2026-10-24 06:58 UTCsource
minimax/minimax-m3.1-flash/sambanova/unknownminimax/minimax-m3.1-flash via SambaNova USD 0.30USD 1.20–1,048,576 2026-10-24 06:58 UTCsource
openai/gpt-oss-120b/sambanova/unknownopenai/gpt-oss-120b via SambaNova USD 0.22USD 0.59–131,072 2026-10-24 06:58 UTCsource