Back/SambaNova

SambaNova

AI & Machine Learning cloud.sambanova.ai

Palo Alto, CA-based AI infrastructure company founded 2017. Private. Operates in AI inference and hardware. Serves enterprises, developers, and data centers with cloud-based AI inference platform using custom AI hardware.

Entry price

Free

Free plan available

Plans tracked

1 add-on

Pricing changes

Latest Wk 41 '26

Pricing model

Usage-based

Free tier available

Pricing Change History

6 changes

20266 changes

Wk 41 '26

Packaging

SambaNova swaps MiniMax-M2.7 for M3, adds $0.06 cached input rate

MiniMax-M2.7 removed from model price table; MiniMax-M3 now takes its row with $0.06 cached input (was N/A), $0.60 input, $2.40 output unchanged. Duplicate M3 row dropped.

View diff

Wk 33 '26

Price increase

SambaNova: gemma-4-31B-it token price increase ($0.22→$0.38 in / $0.59→$1.15 out)

gemma-4-31B-it model pricing increased from $0.22/$0.59 to $0.38/$1.15 per 1M input/output tokens (about 73% and 95% increases respectively)

View diff

Wk 28 '26

Packaging

SambaNova adds cached-token pricing: MiniMax-M2.7 at $0.06 per 1M tokens

Added a new 'Cached Input Tokens' pricing column to the model pricing table; MiniMax-M2.7 is priced at 0.06 USD per 1M cached tokens while all other models show N/A for this metric

View diff

Wk 27 '26

Price decrease

SambaNova cuts gemma-4-31B-it output price 49%, input 42%

gemma-4-31B-it token prices decreased: input from $0.38 to $0.22/1M tokens (42% reduction) and output from $1.15 to $0.59/1M tokens (49% reduction).

View diff

Wk 25 '26

Packaging

SambaNova drops Gemma 3 and Llama 4 Maverick from API lineup

Two models removed from API pricing table: gemma-3-12b-it (was $0.35/$0.59 per 1M tokens) and Llama-4-Maverick-17B-128E-Instruct (was $0.63/$1.80 per 1M tokens).

View diff

Wk 23 '26

Improvement

SambaNova adds Gemma 4 31B at $0.38/$1.15 per million tokens

New model gemma-4-31B-it added to SambaNova pricing table at 0.38 USD/1M input tokens and 1.15 USD/1M output tokens.

View diff