Molmo 2 8B — VRAM, speed & local setup
Written by Jakub Rusinowski · Last updated
An 8.7B vision-language model for images, multi-image input and short video, with pointing and object tracking. Built on Qwen3-8B with a SigLIP 2 vision encoder; Ai2 reports it ahead of other open-weight, open-data models of its class on short video (vendor-reported). 36,864-token context; Apache-2.0.
Molmo 2 8B needs about 6 GB of VRAM at Q4_K_M — quantized weights plus framework overhead, before any KV cache. On Apple Silicon that figure comes out of unified memory.
VRAM and speed by quantization
Quoted against NVIDIA RTX 4090 (24 GB). Includes the KV cache at 8K context, so it reads higher than the headline figure.
| Quant | Memory | VRAM | Speed (est.) | Fit |
|---|---|---|---|---|
| Q2_K 2.63 bpw | 4.9 GB | ~147 tok/s | Fits | |
| Q3_K_M 3.41 bpw | 5.7 GB | ~126 tok/s | Fits | |
| Q4_K_M 4.83 bpw | 7.2 GB | ~100 tok/s | Fits | |
| Q5_K_M 5.67 bpw | 8.1 GB | ~90 tok/s | Fits | |
| Q6_K 6.56 bpw | 9.1 GB | ~80 tok/s | Fits | |
| Q8_0 8.50 bpw | 11.2 GB | ~66 tok/s | Fits | |
| F16 16.00 bpw | 19.3 GB | ~39 tok/s | Fits |
Black marker = usable memory on the NVIDIA RTX 4090 (24 GB). Estimates from the memory-bandwidth roofline on the methodology page. Molmo 2 8B VRAM calculator →
Get Molmo 2 8B running
The cheapest catalogued GPU that runs Molmo 2 8B is the Intel Arc B570 (10 GB).
or compare on Vast.ai from $0.35/hr (typical low · varies)
As an Amazon Associate we earn from qualifying purchases. Cloud GPU links are referral links — we may earn a commission at no extra cost to you.
How to run Molmo 2 8B
Install Ollama, then run:
ollama run molmoSpecifications
Verified — Checked against the primary source — the model card or the vendor spec page — and corroborated by a second independent source.
- Parameters
- 8.7 Billion
- Context window
- 36,864
- Architecture
- Qwen3-8B language model + SigLIP 2 vision encoder
- Provider
- Allen Institute for AI (Ai2)
- Licence
- Apache-2.0
- Availability
- Weights downloadable
- Practical deployment
- Laptop
- Specified at
- Q4_K_M
- System RAM
- 16 GB
- Record updated
- 2026-11-11
Commercial use permitted. No usage restrictions beyond attribution.
Limits and caveats
- Ai2 notes that some training datasets are restricted to academic and non-commercial research; check them before commercial use.
Quality and use cases
Scores as published by the model’s authors or an independent evaluator — quality, not throughput, and not measured by us.
Molmo 2 8B — frequently asked questions
How much VRAM does Molmo 2 8B need?
About 6 GB at Q4_K_M — quantized weights plus framework overhead, before any KV cache. The cache grows with context length and is added on top; the table above folds it in. Apple Silicon counts unified memory toward the same figure.
Does Molmo 2 8B run on an RTX 4090 (24 GB)?
Yes. Molmo 2 8B needs about 6 GB at Q4_K_M, inside a 24 GB card, at an estimated 100 tokens/sec.
How do I run Molmo 2 8B locally?
Install Ollama and run `ollama run molmo`. That pulls the weights and starts a local OpenAI-compatible endpoint; after the download nothing leaves the machine.