Over recent months, we have intensified our focus on developing foundation models that deliver exceptional utility and performance.
S5-nano
maritaca-ai·Instruct·released 2 Nov 2026
ollama run hf.co/maritaca-ai/S5-nano-GGUF:Q4_K_MS5-nano is a 4B-parameter model from Maritaca AI. It has two modes:
How much VRAM does S5-nano need?
At Q4_K_M S5-nano needs about 4.3 GB with an 8K context (4.3 GB at Q4, 11 GB at full precision). Pick your GPU to see what fits and how fast it runs.
Needs: ≈4.2 GB · Q4_K_M · 8K context — Pick a GPU to check the fit.
Advanced: context 8K · KV cache 16-bit
Smallest GPU for each quant and context (KV cache 16-bit)
| Quantisation | 8K | 16K | 32K | 64K | 128K | 256K |
|---|---|---|---|---|---|---|
| Q3_K_M | ||||||
| Q4_K_M | ||||||
| Q5_K_M | ||||||
| Q6_K | ||||||
| Q8_0 | ||||||
| FP16 |
≤16 GB card24–32 GB48–96 GBmore
| Quantisation | Total |
|---|---|
| Q2_KHeavy lossSmallest, noticeably worse quality | ≈3.3 GB |
| Q3_K_MNoticeable lossSmall, some quality loss | ≈3.6 GB |
| Q4_K_MrecommendedSmall lossRecommended default for local use | ≈4.2 GB |
| Q5_K_MNear-losslessNear-lossless for most tasks | ≈4.8 GB |
| Q6_KNear-losslessPractically lossless | ≈5.2 GB |
| Q8_0LosslessLossless in practice, twice the size of Q4 | ≈6.0 GB |
| FP16 / BF16LosslessFull precision, as released | ≈10.0 GB |
Estimates for 8K tokens of context: weights + KV cache + runtime overhead. Speed assumes memory-bound decoding at your GPU’s bandwidth. Pick a GPU above and the fit column checks every row against it. Weights marked “file” use the real GGUF download size; “est.” rows are computed from the parameter count.
How to pick a quant
Take the biggest quant that fits with room left for context. Q4_K_M is the usual sweet spot. Below Q4 quality drops fast, above Q6 you mostly pay memory for nothing. Big models (70B+, large MoE) tolerate low quants better than small ones.
- LosslessF16, Q8_0
- Near-losslessQ6_K, Q5_K_M
- Small lossQ5_K_S, Q4_K_M, IQ4_NL, IQ4_XS, Q4_K_S
- Noticeable lossQ4_0, Q3_K_L, Q3_K_M, IQ3_M
- Heavy lossIQ3_XXS, Q3_K_S, IQ2_M, Q2_K
Typical quality loss, not measured per model. Tiers come from KL divergence against FP16 on Llama 3 8B: llama.cpp perplexity README: LLaMA 3 8B scoreboard (KLD and ΔPPL vs FP16, Wikitext-2); llama.cpp quantize tool: ΔPPL per quant type @ Llama-3-8B
How much context fits?
At 8K context, Q4_K_M needs ≈4.2 GB(2.6 GB weights + 1.0 GB KV cache + 0.6 GB overhead). Pick a GPU above to see how much context fits.
Weights + overheadKV cache
Can the RTX 4060 run S5-nano? →
Speed vs quality per quant
Smaller quants run faster and fit in less memory, but lose quality.
Fits in VRAMTight fitCPU offload✕ won't runQuality is typical KL divergence vs FP16, not measured for this model.
Real-world speed of S5-nano
Tokens per second measured by people running it at home. A GPU shows a median once 3 reports agree on the same setup. Reports far from our estimate wait for a review. Measure yours with llama-bench →
No medians yet. Be one of the first 3 to report a setup.
All reports
Run S5-nano locally
--quantQ4_K_Mdefault; pick your GPU to tailor itmaritaca-ai/S5-nano-GGUF
ollama run hf.co/maritaca-ai/S5-nano-GGUF:Q4_K_M llama-server -hf maritaca-ai/S5-nano-GGUF:Q4_K_MDownloads the GGUF on first run. Installed through llama.app? The same flags work with `llama serve`.
lms get maritaca-ai/S5-nano-GGUF@Q4_K_MThen load it from the LM Studio app or with `lms load`.
vllm serve maritaca-ai/S5-nano vllm serve maritaca-ai/S5-nano-FP8services:
vllm:
image: vllm/vllm-openai:latest
command: maritaca-ai/S5-nano
ports:
- "8000:8000"
ipc: host
volumes:
- ~/.cache/huggingface:/root/.cache/huggingface
deploy:
resources:
reservations:
devices:
- driver: nvidia
count: all
capabilities: [gpu]API on port 8000.
sglang serve maritaca-ai/S5-nano --host 0.0.0.0 --port 30000services:
sglang:
image: lmsysorg/sglang:latest
command: sglang serve maritaca-ai/S5-nano --host 0.0.0.0 --port 30000
ports:
- "30000:30000"
ipc: host
volumes:
- ~/.cache/huggingface:/root/.cache/huggingface
deploy:
resources:
reservations:
devices:
- driver: nvidia
count: all
capabilities: [gpu]API on port 30000.
from transformers import pipeline
pipe = pipeline("image-text-to-text", model="maritaca-ai/S5-nano", dtype="auto", device_map="auto")
messages = [
{
"role": "user",
"content": [
{"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/bee.jpg"},
{"type": "text", "text": "Describe this image."},
],
}
]
out = pipe(text=messages, max_new_tokens=256)
print(out[0]["generated_text"][-1]["content"])Quantised versions and downloads
| Repository | Format | Quant | Size |
|---|---|---|---|
| maritaca-ai/S5-nano-FP8 | Safetensors | FP8 | — |
| bartowski/maritaca-ai_S5-nano-GGUF | GGUF | IQ2_M | 1.7 GB |
| bartowski/maritaca-ai_S5-nano-GGUF | GGUF | Q2_K | 1.8 GB |
| bartowski/maritaca-ai_S5-nano-GGUF | GGUF | IQ3_XXS | 2.0 GB |
| bartowski/maritaca-ai_S5-nano-GGUF | GGUF | Q3_K_S | 2.0 GB |
| bartowski/maritaca-ai_S5-nano-GGUF | GGUF | IQ3_XS | 2.1 GB |
| bartowski/maritaca-ai_S5-nano-GGUF | GGUF | Q3_K_M | 2.1 GB |
| bartowski/maritaca-ai_S5-nano-GGUF | GGUF | Q3_K_L | 2.2 GB |
| bartowski/maritaca-ai_S5-nano-GGUF | GGUF | IQ3_M | 2.3 GB |
| bartowski/maritaca-ai_S5-nano-GGUF | GGUF | IQ4_XS | 2.5 GB |
About S5-nano
S5-nano is a 4B-parameter model from Maritaca AI. It has two modes: • Think: reasons before answering (the default mode). • Instruct: answers directly, without a reasoning step. The context window supports up to 262,144 tokens, and the weights include an MTP (multi-token prediction) layer for speculative decoding. FP8 is recommended for GPUs with native FP8 support (Ada, Hopper, and Blackwell). The GGUF repository provides BF16, Q8_0, and Q4_K_M variants. Explore the results in Observatório Maritaca Results for the original S5-nano checkpoint, evaluated in BF16 on text tasks. Each primary metric is weighted equally.
Full model card on Hugging Face →Alternatives to S5-nano
All alternatives →Meet Qwen3-VL — the most powerful vision-language model in the Qwen series to date.
Over recent months, we have intensified our focus on developing foundation models that deliver exceptional utility and performance.
WeVisOne-VL (WeChat Vision All-in-One) is a unified vision-language model developed by WeChat Vision, Tencent Inc. Built upon Qwen3-VL, it unifies object-level perception (open-vocabulary object detection, referring localization, object counting, object pointing, and OCR) and pixel-level perception…
H2O-Lightning-4B is a 4B-parameter decision model from H2O.ai, built on Qwen/Qwen3.5-4B. It answers typed decision questions about a record (a document, a ticket, a policy, a conversation) and about images that come with the record (photos, screenshots, scanned documents, charts).
* License: The use of MedGemma is governed by the Health AI Developer Foundations terms of use. MedGemma has not been evaluated or optimized for multi-turn applications.
S5-nano FAQ
- How much VRAM does S5-nano need?
- About 4.3 GB at Q4_K_M with an 8K context. Q8_0 needs about 6.3 GB and full precision about 11 GB. Longer contexts need more memory for the KV cache.
- Can S5-nano run on the RTX 4060?
- Yes. It fits at Q8_0 (about 6.0 GB), at an estimated 30 tokens/s. It fits comfortably on the Apple M4 (16 GB), Radeon RX 6700 XT, RTX 3060 12GB.
- Can I use S5-nano commercially?
- Yes. It is released under Apache 2.0. Permissive: use, modify and sell freely; keep the notice and licence text.
- What is the context length of S5-nano?
- 262,144 tokens (256K). The full window needs far more memory than an 8K context.
- How do I run S5-nano with Ollama or llama.cpp?
- With Ollama: ollama run hf.co/maritaca-ai/S5-nano-GGUF:Q4_K_M. With llama.cpp: llama-server -hf maritaca-ai/S5-nano-GGUF:Q4_K_M, which starts an OpenAI-compatible server.
Rate this model
Data from Hugging Face, the model card and public leaderboards. Memory and speed figures are estimates; real numbers depend on the runtime, context length and drivers. Always check the licence yourself before commercial use.
0 comments