SambaNova

Data handling can vary by model and deployment. The privacy badge above shows the more conservative of the provider-wide fallback and the policies for routes on this page; each model page shows the policy for that specific route.

Models served by SambaNova

Model name

Model	Context	Input /M	Output /M	Cache read /M	Cache hit	TTFT	Throughput
DeepSeek V3.1 FP8 deepseek-ai/DeepSeek-V3.1:sambanova	131k	$0.682/M	$1.57/M	N/A	N/A	2.7s	81 tps
DeepSeek V3.2 deepseek/deepseek-v3.2:sambanova	33k	$3.15/M	$4.72/M	N/A	N/A	2s	28.5 tps
DeepSeek V3.2 (Thinking) deepseek/deepseek-v3.2:thinking:sambanova	33k	$3.15/M	$4.73/M	N/A	N/A	2s	28.5 tps
Gemma 4 31B google/gemma-4-31b-it:sambanova	131k	$0.399/M	$1.21/M	N/A	0%	3.7s	107 tps
Gemma 4 31B (Thinking) google/gemma-4-31b-it:thinking:sambanova	131k	$0.399/M	$1.21/M	N/A	0%	3.7s	107 tps
GPT OSS 120B openai/gpt-oss-120b:sambanova	131k	$0.147/M	$0.998/M	N/A	N/A	1s	342 tps
Llama 3.3 70B Instruct meta-llama/llama-3.3-70b-instruct:sambanova	131k	$0.472/M	$0.945/M	N/A	N/A	0.8s	148 tps
MiniMax M3.1 Flash minimax/minimax-m3.1-flash:sambanova	1.0M	$0.315/M	$1.26/M	$0.0315/M	49.2%	13.2s	176.7 tps

DeepSeek V3.1

FP8

deepseek-ai/DeepSeek-V3.1:sambanova

Context

131k

Input /M

$0.682/M

Output /M

$1.57/M

Cache read /M

N/A

Cache hit

N/A

TTFT

2.7s

Throughput

81 tps

DeepSeek V3.2

deepseek/deepseek-v3.2:sambanova

Context

33k

Input /M

$3.15/M

Output /M

$4.72/M

Cache read /M

N/A

Cache hit

N/A

TTFT

2s

Throughput

28.5 tps

DeepSeek V3.2 (Thinking)

deepseek/deepseek-v3.2:thinking:sambanova

Context

33k

Input /M

$3.15/M

Output /M

$4.73/M

Cache read /M

N/A

Cache hit

N/A

TTFT

2s

Throughput

28.5 tps

Gemma 4 31B

google/gemma-4-31b-it:sambanova

Context

131k

Input /M

$0.399/M

Output /M

$1.21/M

Cache read /M

N/A

Cache hit

0%

TTFT

3.7s

Throughput

107 tps

Gemma 4 31B (Thinking)

google/gemma-4-31b-it:thinking:sambanova

Context

131k

Input /M

$0.399/M

Output /M

$1.21/M

Cache read /M

N/A

Cache hit

0%

TTFT

3.7s

Throughput

107 tps

GPT OSS 120B

openai/gpt-oss-120b:sambanova

Context

131k

Input /M

$0.147/M

Output /M

$0.998/M

Cache read /M

N/A

Cache hit

N/A

TTFT

1s

Throughput

342 tps

Llama 3.3 70B Instruct

meta-llama/llama-3.3-70b-instruct:sambanova

Context

131k

Input /M

$0.472/M

Output /M

$0.945/M

Cache read /M

N/A

Cache hit

N/A

TTFT

0.8s

Throughput

148 tps

MiniMax M3.1 Flash

minimax/minimax-m3.1-flash:sambanova

Context

1.0M

Input /M

$0.315/M

Output /M

$1.26/M

Cache read /M

$0.0315/M

Cache hit

49.2%

TTFT

13.2s

Throughput

176.7 tps

Prices are per million tokens. TTFT is the observed time to first token. Selectable routes include the applicable provider-selection markup; fixed routes show the standard model price. Availability and pricing refresh continuously; each model page shows the live provider comparison.

Using SambaNova via the API

Call a model served by SambaNova using its displayed model ID, including the provider suffix when selection is supported.

curl https://nano-gpt.com/api/v1/chat/completions \
  -H "Authorization: Bearer $NANOGPT_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "deepseek-ai/DeepSeek-V3.1:sambanova",
    "messages": [{"role": "user", "content": "Hello"}]
  }'

For models with provider selection, append :sambanova to the model ID, set "provider": "sambanova" in the request body, or send an X-Provider: sambanova header. Explicit provider selection adds a route-specific markup over that provider's base price; the prices in the table above already include it. Models without provider selection use the displayed model ID without a provider suffix or selection surcharge.

SambaNova offers multiple service lanes, each pinned with its own id: sambanova, sambanova-high-throughput (High Throughput). The table shows which lane serves each row.

See the API documentation for provider preferences, price-aware routing, and error behavior.