North Micro Vision Instruct
North Micro Vision Instruct is a 2.4B-parameter open-weight, Apache 2.0-licensed vision-language model from CohereLabs for the image-text-to-text pipeline. It accepts interleaved text and images and outputs text, using a 2B-parameter language model, a 400M-parameter custom vision encoder based on SigLIP 2 SO400M, and bfloat16 checkpoint precision. It supports native-resolution image processing and multilingual, multi-image input covering English, German, French, Spanish, Italian, Portuguese, Hindi, Japanese, Korean, Chinese, Arabic, and more. Its language backbone has a 128K-token context window, while multimodal training context and the validated operating range for multimodal prompts are up to 8K tokens. The model is Apache 2.0 licensed and intended for prototyping, task-specific fine-tuning, and specialized multimodal applications.
At a glance
- Parameters
- 2.48B
- Architecture
- cohere_compass
- License
- apache-2.0
- Software
- Transformers
Will it run on your machine?
Save your machine to see a personalized rating and its reasoning.
Add my hardwareBest for
- Image question answering and visual question answering
- Image captioning
- Visual grounding
- OCR on images and documents
- Chart and document understanding
Tradeoffs
- Multimodal prompts are validated only up to 8K tokens; longer multimodal contexts may rely on extrapolation and have not been benchmarked.
- The source does not specify hardware requirements or benchmarked performance.
- Compatibility is stated only for Transformers 5.16.0; other versions are not addressed.
- The checkpoint precision is stated as bfloat16; other precisions are not stated.
- The model card describes broad multilingual coverage but does not enumerate all supported languages.
Ways to run it
Transformers · See official guide
Repository-specific command in publisher documentation; confirm dependencies and hardware in the source.
import torch
from transformers import AutoModelForImageTextToText, AutoProcessor
model_id = "CohereLabs/North-Micro-Vision-Instruct"
processor = AutoProcessor.from_pretrained(
model_id,
)
model = AutoModelForImageTextToText.from_pretrained(
model_id,
dtype="auto",
device_map="auto",
)
# To enable Flash Attention 2, load the model with the following settings:
# model = AutoModelForImageTextToText.from_pretrained(
# model_id,
# dtype=torch.bfloat16,
# attn_implementation="flash_attention_2",
# device_map="auto",
# )
image_url = "https://cdn-uploads.huggingface.co/production/uploads/66d732effe6684fc16b12c28/Io_5OCmftsmH-n158ZtPs.png"
messages = [
{
"role": "user",
"content": [
{"type": "image", "url": image_url},
{"type": "text", "text": "What do you see?"},
],
}
]
inputs = processor.apply_chat_template(
messages,
tokenize=True,
add_generation_prompt=True,
return_tensors="pt",
return_dict=True,
).to(model.device)
outputs = model.generate(
**inputs,
max_new_tokens=128,
do_sample=True,
temperature=0.7,
top_p=0.8,
top_k=20,
)
generated_ids = [
output_ids[len(input_ids) :]
for input_ids, output_ids in zip(inputs.input_ids, outputs)
]
response = processor.batch_decode(
generated_ids,
skip_special_tokens=True,
clean_up_tokenization_spaces=False,
)[0]
print(response)Explore its uses
Keep exploring
Bonsai 2 · 27B
A compressed 27B-class reasoning model with publisher-provided GGUF packs and a dedicated llama.cpp fork for CUDA, Metal, and CPU.
DeepSeek V4.1 Flash
A multimodal reasoning model with a compressed key-value cache, published as open weights.
Qwen3.8-Flash-Next
An experimental open-weight multimodal model with sparse attention and 262K native context.