// Cohere Command / Aya · Vision-language
North-Micro-Vision-Instruct: VRAM & local deployment
2.48B parameters. About 2.7 GB at 4-bit and 6.6 GB at BF16 with one 8K-token request.
released 2026-08-10licence: Commercial OKmemory figures are estimates
01Specs
| Publisher | Cohere · CohereLabs |
|---|---|
| Family / series | Cohere Command / Aya · North-Micro-Vision |
| Type | Vision-language (image-text-to-text) |
| Parameters | 2.48B Source: counted from official safetensors weights |
| Context length | 488K tokens (500,000)max_position_embeddings in config.json |
| Architecture | 28 layers (7 full-attention) · 21 sliding-window layers (4096) · GQA |
| Released weights | BF16 · 5.0 GBdownload |
| Licence | Apache-2.0 · Commercial OKPermissive licence (Apache-2.0 / MIT / CC-BY / OpenMDW): commercial use allowed. |
| Released | 2026-08-10 (Hugging Face repo created) |
02How much VRAM it needs (estimate)
Total = weights + 10% + 0.5GB overhead + KV cache for one request (FP16 KV). More concurrent users multiply the KV column.
| Precision | Weights | Total @ 8K | Total @ 32K | Total @ 128K | Smallest GPU | Mac |
|---|---|---|---|---|---|---|
| BF16 / FP16 (original) | 5.0 GB | 6.6 GB | 7.3 GB | 10.1 GB | 8GB GPU | Mac 16GB |
| FP8 / INT8 / Q8_0 | 2.6 GB | 4.0 GB | 4.7 GB | 7.5 GB | 8GB GPU | Mac 16GB |
| 4-bit (Q4_K_M / AWQ / GPTQ-Int4) | 1.5 GB | 2.8 GB | 3.5 GB | 6.3 GB | 8GB GPU | Mac 16GB |
KV cache per request: 8K 0.6 GB · 32K 1.3 GB · 128K 4.1 GB
Bits per weight: BF16 16, FP8 8.5, 4-bit 4.85. “Fits” = total ≤ 90% of memory; Mac GPU can use ~2/3 of unified memory (≤36GB) or ~3/4 (larger). For MoE, all experts must be in memory, so size by total parameters. Method →
03How to run it
Only runtimes the publisher or the runtime’s official library names for this model.
- vLLM — mentioned in the official model card
- MLX — mentioned in the official model card
- Transformers — mentioned in the official model card
04Official repositories & quantised builds
1 official repo(s) from CohereLabs.
- North-Micro-Vision-Instruct Instruct
05Other Cohere Command / Aya models
All 14 →| Model | Params | 4-bit total | Released |
|---|---|---|---|
| North-Mini-Code-1.0 | 30B / 3B | 21.0 GB | 2026-06-05 |
| command-a-plus-05-2026-bf16 | 218B / 25B | 147 GB | 2026-05-11 |
| tiny-aya-base | 3.35B | 2.7 GB | 2026-02-13 |
| tiny-aya-global | 3.35B | 2.7 GB | 2026-02-13 |
| command-a-reasoning-08-2025 | 111.1B | 74.6 GB | 2025-08-12 |
| command-a-vision-07-2025 | 111.9B | 75.1 GB | 2025-07-28 |
| c4ai-command-a-03-2025 | 111.1B | 74.6 GB | 2025-03-11 |
| aya-vision-8b | 8.63B | 6.3 GB | 2025-03-02 |
06Sources
Data checked 2026-09-30. Parameters, context and licence are copied from the official repository; memory figures are our estimates.