Skip to main content
Source: SambaNova model docs. All model IDs use the prefix sambanova/. Powered by SambaNova RDU (Reconfigurable Dataflow Unit) chips designed for large-scale AI inference.
SambaNova models are not yet available through the Timbal platform proxy. A SAMBANOVA_API_KEY is required to use these models. If you’d like to access SambaNova models via your Timbal API key, please contact sales.

​
DeepSeek

DeepSeek-V3.2

Reasoning · Speedsambanova/DeepSeek-V3.2DeepSeek V3.2 running on SambaNova RDU hardware.
  • $3.00 / $4.50
  • 8K context
  • Text input

DeepSeek-V3.1

Reasoning · Speedsambanova/DeepSeek-V3.1DeepSeek V3.1 running on SambaNova RDU hardware.
  • $3.00 / $4.50
  • 128K context
  • Text input

​
Meta Llama

Llama-4-Maverick-17B-128E-Instruct

Reasoning · Speedsambanova/Llama-4-Maverick-17B-128E-InstructMeta Llama 4 Maverick with 17B active parameters across 128 experts. Supports image input.
  • $0.63 / $1.80
  • 128K context
  • Text, Image input

Meta-Llama-3.3-70B-Instruct

Reasoning · Speedsambanova/Meta-Llama-3.3-70B-InstructMeta Llama 3.3 70B instruction-tuned model running on SambaNova RDU hardware.
  • $0.60 / $1.20
  • 128K context
  • Text input

​
Google Gemma

gemma-4-31B-it

Reasoning · Speedsambanova/gemma-4-31B-itGoogle Gemma 4 31B instruction-tuned model on SambaNova RDU hardware.
  • $0.40 / $0.80
  • 128K context
  • Text, Image input

gemma-3-12b-it

Reasoning · Speedsambanova/gemma-3-12b-itGoogle Gemma 3 12B instruction-tuned model on SambaNova RDU hardware.
  • $0.10 / $0.20
  • 128K context
  • Text input

​
OpenAI OSS / MiniMax

gpt-oss-120b

Reasoning · Speedsambanova/gpt-oss-120bOpenAI’s open-weight 120B MoE on SambaNova RDU hardware. Near-parity with o4-mini on reasoning benchmarks. Apache 2.0.
  • $0.22 / $0.59
  • 128K context
  • Text input

MiniMax-M2.7

Reasoning · Speedsambanova/MiniMax-M2.7MiniMax M2.7 large-scale MoE running on SambaNova RDU hardware.
  • $0.30 / $1.20
  • 128K context
  • Text input