Available Models

Sutro currently offers access to the category leading open-source language models. All models are currently run at full precision (no quantization). We will continue to add more models with increasing modalities, capabilities, and task-specific training over time.

Text-to-Text Models

Model Name	Model ID	Context Window (input tokens)	Modalities	Price (USD/Million tokens, input and output)
Llama 3.2 3B Instruct	llama-3.2-3b	128k	Text	0.007
Llama 3.1 8B Instruct	llama-3.1-8b	128k	Text	0.013
Llama 3.3 70B Instruct	llama-3.3-70b-8k	8k	Text	0.148
Llama 3.3 70B Instruct	llama-3.3-70b-64k	64k	Text	0.149
Gemma 3 4B Instruct	gemma-3-4b-it	128k	Text	0.01
Gemma 3 27B Instruct	gemma-3-27b-it-16k	16k	Text	0.059
Gemma 3 27B Instruct	gemma-3-27b-it-128k	128k	Text	0.08
Qwen 3 4B Instruct	qwen-3-4b	32k	Text	0.059
Qwen 3 32B Instruct	qwen-3-32b	32k	Text	0.08

Reasoning Models (text-to-text)

Model Name	Model ID	Context Window (input tokens)	Modalities	Price (USD/Million tokens, input and output)
Qwen 3 4B Instruct	qwen-3-4b-thinking	32k	Text	0.04
Qwen 3 32B Instruct	qwen-3-32b-thinking	32k	Text	0.08

Embedding Models

Model Name	Model ID	Context Window (input tokens)	Embedding Dimension	Modalities	Price (USD/Million tokens, input and output)
Multilingual e5 Large Instruct	multilingual-e5-large-instruct	512	1024	Text	0.0001
GTE Qwen 2 7B Instruct	gte-qwen2-7b-instruct	32k	3584	Text	0.007

Custom Models

We also support custom and fine-tuned models. Please contact team@sutro.sh for more information.