MA

Marco o1

MarcoGGUF by bartowskiUpdated 2024-11-22
Download on Hugging Face
Parameters7.6B
Context32K
Layers28
Architectureqwen2
Smallest quant2.8 GB
Quick answer

Marco o1 needs about 5.7 GB of memory at Q4_K_M (a 4.7 GB file) with an 8K context window, so it fits on a 6 GB GPU. This repository has 25 quants, from 2.8 GB (IQ2_M) to 30.5 GB (F32). Longer context needs more: at 32K tokens the KV cache alone adds about 1.9 GB.

Marco o1 – 7.6B parameter LLM focused on open‑ended reasoning

Marco o1 is a 7.6B parameter LLM built on the qwen2 architecture. It is fine‑tuned with chain‑of‑thought data, Monte Carlo Tree Search, and reflection mechanisms for complex problem solving. The model is released under the Apache‑2.0 license.

Good at

  • Open‑ended reasoning
  • Complex problem solving
  • Chain‑of‑thought generation

Limits

  • No vision support
  • Requires 32‑bit floating point for full precision
  • Large VRAM needed for higher quants

Best for

  • Research on reasoning models
  • Developing reasoning‑based applications
  • Testing MCTS integration

Who it’s for. Researchers and developers interested in advanced reasoning capabilities. The model is suitable for those who need a large language model that can handle open‑ended tasks and experiment with chain‑of‑thought and MCTS techniques.

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
IQ2_M2.8 GB 3.8 GB
Fits Get file →
Q2_K3.0 GB 4.0 GB
Fits Get file →
IQ3_XS3.3 GB 4.3 GB
Fits Get file →
Q3_K_S3.5 GB 4.5 GB
Fits Get file →
Q2_K_L3.5 GB 4.5 GB
Fits Get file →
IQ3_M3.6 GB 4.5 GB
Fits Get file →
Q3_K_M3.8 GB 4.8 GB
Fits Get file →
Q3_K_L4.1 GB 5.1 GB
Fits Get file →
IQ4_XS4.2 GB 5.2 GB
Fits Get file →
Q4_04.4 GB 5.4 GB
Fits Get file →
Q4_04.4 GB 5.4 GB
Fits Get file →
Q4_04.4 GB 5.4 GB
Fits Get file →
Q4_04.4 GB 5.4 GB
Fits Get file →
Q4_K_S4.5 GB 5.4 GB
Fits Get file →
Q3_K_XL4.6 GB 5.5 GB
Fits Get file →
Q4_K_M4.7 GB 5.7 GB
Fits Get file →
Q4_K_L5.1 GB 6.1 GB
Fits Get file →
Q5_K_S5.3 GB 6.3 GB
Fits Get file →
Q5_K_M5.4 GB 6.4 GB
Fits Get file →
Q5_K_L5.8 GB 6.8 GB
Fits Get file →
Q6_K6.3 GB 7.2 GB
Tight Get file →
Q6_K_L6.5 GB 7.5 GB
Tight Get file →
Q8_08.1 GB 9.1 GB
Too big Get file →
F1615.2 GB 16.2 GB
Too big Get file →
F3230.5 GB 31.4 GB
Too big Get file →

For 8 GB VRAM, use Q4_0 (file size 4.43 GB, estimated memory 5.4 GB). For 12 GB VRAM, Q5_K_M (5.44 GB, 6.4 GB) or Q4_K_M (4.68 GB, 5.7 GB) are suitable. For 16 GB VRAM, Q6_K (6.25 GB, 7.2 GB) or Q5_K_L (5.78 GB, 6.8 GB) provide higher quality.

What GPU do you need for Marco o1?

These figures add the file size, an fp16 KV cache at an 8K context window (worked out from this model’s own header) and about 0.5 GB for the runtime. They are estimates, so leave a little headroom, and expect to use less context or a smaller quant if you are right at the limit.

  • 4 GB VRAM: the best quant that fits is Q2_K (3.0 GB file, about 4.0 GB in use).
  • 6 GB VRAM: the best quant that fits is Q4_K_M (4.7 GB file, about 5.7 GB in use).
  • 8 GB VRAM: the best quant that fits is Q6_K_L (6.5 GB file, about 7.5 GB in use).
  • 12 GB VRAM: the best quant that fits is Q8_0 (8.1 GB file, about 9.1 GB in use).
  • 16 GB VRAM: the best quant that fits is Q8_0 (8.1 GB file, about 9.1 GB in use).
  • 24 GB VRAM: the best quant that fits is F16 (15.2 GB file, about 16.2 GB in use).
  • 32 GB VRAM: the best quant that fits is F32 (30.5 GB file, about 31.4 GB in use).

Which Marco o1 quant should you choose?

Q4_K_M is the usual starting point for Marco o1: it balances file size and quality well. Here it is a 4.7 GB file.

The Q8_0 file is 8.1 GB, about 1.7 times the size of Q4_K_M. It keeps the weights closer to the original, which only matters if you have the memory to spare.

If memory is tight, the smallest file is IQ2_M at 2.8 GB, with the quality trade-offs that come with fewer bits. Use it only when memory is very tight. Expect a clear drop in quality, which is usually worse on small models and easier to live with on large ones.

New to quant names? Read GGUF quantization explained.

Specifications

Architecture from the GGUF header

Architectureqwen2
Parameters7.6B
Layers28
Trained context32,768 tokens
KV cache (fp16)56 KB per token

Source

Quantized bybartowski
FamilyMarco
Licenseapache-2.0
Downloads (HF)1,933
Last updated2024-11-22

Frequently asked questions about Marco o1

How much VRAM does Marco o1 need?

About 5.7 GB at Q4_K_M with an 8K context window. That is the 4.7 GB file, plus the KV cache and roughly 0.5 GB of runtime overhead. Smaller quants need less and longer context needs more.

Can I run Marco o1 on an 8 GB GPU?

Yes, with Q6_K_L: the 6.5 GB file needs about 7.5 GB at 8K context. Larger quants in this repository do not fit in 8 GB.

What is the file size of Marco o1 GGUF?

It depends on the quant: IQ2_M is 2.8 GB, Q2_K is 3.0 GB, IQ3_XS is 3.3 GB, Q3_K_S is 3.5 GB, Q2_K_L is 3.5 GB, IQ3_M is 3.6 GB and 19 more.

What context length does Marco o1 support?

The GGUF header lists a trained context length of 32,768 tokens (32K). The KV cache grows with the context you use: about 0.5 GB at 8K and 1.9 GB at 32K.

Where can I download Marco o1 GGUF?

The files are on Hugging Face in the bartowski/Marco-o1-GGUF repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.

What license does Marco o1 use?

Hugging Face lists the license as apache-2.0. Licenses can differ between the quantized files and the original model, so check the repository and the base model before commercial use.

Can I run Marco o1 on 8 GB VRAM?

Yes, the Q4_0 quant (4.43 GB file) requires about 5.4 GB of VRAM at 8 k context, which fits within an 8 GB budget.

Does Marco o1 support vision tasks?

No, the model does not include vision capabilities.

Other GGUFs of Marco-o1

Similar models

Browse all →

More from bartowski

All by bartowski →

More Marco models

All Marco →

Other small models

All small models →