Skip to content
Local AI Wizard
Your machine
Theme

North-Micro-Vision-Instruct

Cohere · CohereLabs/North-Micro-Vision-Instruct

Weights license: Apache 2.0 · code and components may use other terms

Reported by the publisher22 resultsCohere's model card · read 2026-10-25 · not ranked · see them
Memory needed~2 GBQ4 weights + working room, before the machine's reserve
Parameters2.5 BHub safetensors total
Repository updated2026-09-30A commit is not a release date

Quality

No comparable score

We track no public, comparable benchmark that covers this model yet. No score is inferred.

Reported by the publisher

22 results · Cohere's model card · read 2026-10-25
MMBench DEV_EN_V11
0.687
MMStar
0.518
RealWorldQA
0.622
GQA TestDev_Balanced
0.574
MTL MMBench_DEV
0.636
MMMB
0.728
BLINK
0.527
ChartQA Test
0.808
Every reported result22
DocVQA VAL
0.921
InfoVQA VAL
0.652
OCRBench v2_en
0.367
OCRBench
0.792
AI2D_TEST
0.775
CharXiv DQ
0.600
MMMU DEV_VAL
0.329
RefCOCO avg †
0.732
CountBench
0.725
HallusionBench
0.615
MMLU test
0.504
MMLU-Pro test
0.307
Multi-If
0.373
IFEval
0.749

Values as printed on the card, with the publisher's own settings (card section: Benchmark Results); column “North-Micro-Vision-Instruct”. Not comparable across publishers and not used in any ranking. Model card

Runs on

Fast memory holds it 119

The planning estimate fits the memory the model can use quickly: graphics card VRAM, or unified memory within the macOS GPU share.

Show all 119

Hardware profiles grouped by the Q4 planning estimate on each machine (dedicated GPUs reserve 2 GB; unified and system memory 25%, 2 to 8 GB): the fast memory holds it, it is within memory but outside the fast part (layers offloaded to PC RAM, or above the macOS GPU share), or the machine is one memory tier short. Not a measured fit or speed.

Consider it for

Cohere released North-Micro-Vision-Instruct on 2026-08-12: a vision-language model that answers about images, with 2.48B parameters on the Hugging Face count and the Apache 2.0 license. Languages the publisher declares: en, de, fr, es, it, pt, hi, ja, ko, zh, ar.

Check first

License: commercial use allowed. At Q4 the weights load about 1.9 GB before each machine's reserve; image resolution, video frames and context add memory, and the vision encoder needs support in the runtime you pick.

Measured speed

No published measurement for this checkpoint

Tokens per second are shown only when a source publishes them with hardware, build and setup. They are never estimated.

How to run

Runtime compatibility and tokens per second have not been measured for this entry. Published metadata and a memory estimate do not prove a successful local run.

Library in source: transformers

Documented ways to run it

Copied as printed in the linked source (read 2026-10-18); not run by this site.

Setup stated first by the source · Transformers
uv pip install accelerate pillow "transformers==5.16.0"
Transformers · model card · Transformers
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)

Choose my computer

Technical details

What the source and the catalog store for this checkpoint. Missing fields are left out, never zero.

Task
image-text-to-text
Architecture
CohereCompassForConditionalGeneration
Library in source
transformers
Weight license
Apache 2.0CohereLabs/North-Micro-Vision-Instruct model card (Hugging Face: "released under the Apache 2.0 license") · read 2026-10-19source
Commercial use
Allowed Apache License 2.0 text (SPDX) · read 2026-10-18license text
Languages
en, de, fr, es, it, pt, hi, ja, ko, zh, arModel card metadata · read 2026-10-25source
Released
Meet North Micro Vision: A 2.4B Native-Resolution Vision-Language Model (Hugging Face blog post by the publisher; JSON-LD datePublished) · read 2026-10-19secondary source
Every stored field12 fields
Revision checked
fa548d9aa62c1220bfd21fc6cfc465f1bd2a1a40
Architecture
CohereCompassForConditionalGeneration
Library
transformers
Hub task tag
image-text-to-text
Formats
transformers, safetensors
Access
Open download
Publisher kind
Official
Hub parameter count
2,484,847,856safetensors.total
Repository created
2026-08-10creation is not the model's release
Repository updated
2026-09-30a commit is not a release date
Hub downloads (30 days) · likes
35,947 · 154this repository only · read 2026-10-25
Catalog state
Hub config or weights found; no inventory benchmark link

Sources

Identity and published specifications come from the checkpoint page. Benchmark scores belong to the named release and do not measure speed on your computer.

Curated facts checked: 2026-10-19. The license and release date cite the publisher's sources; parameters, files and task come from the Hugging Face record, and memory is a labelled estimate from its parameter count. A value no source states stays empty; a repository commit is not a model release date.