GLM-5.3 is live. Run Z.ai's latest model on Friendli Model APIs. Try it today ➜
maritaca-ai
Available on FriendliAI
Run this model inference on single tenant GPU with unmatched speed and reliability at scale.
Model Details
Model Provider
maritaca-ai
Model Tree
Input Modalities
Output Modalities
Supported Functionality
| Version | Format | Size | For |
|---|---|---|---|
| S5-nano | BF16 | 8.7 GB | vLLM, SGLang |
| S5-nano-FP8 | FP8 | 5.7 GB | vLLM, SGLang |
| S5-nano-GGUF | GGUF | 2.8 to 8.7 GB | llama.cpp |
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.
The FP8 version uses E4M3 weights for linear layers, with one scale per 128×128 block, and dynamically quantizes activations during inference. Embeddings, the vision encoder, and some small projections remain in BF16. The MTP layer is included.
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.
Total, active, and effective parameter counts measure different things. LatamGPT 70B is a larger reference model; LFM2.5 has approximately 24B total parameters and 2B active parameters.
Bold indicates the best result within each group with a documented evaluation protocol, in the same mode. Different groups may produce more than one highlighted result in the same row. † indicates that the model produced reasoning in Instruct mode even though it was disabled in the request.
Sampling parameters match the mode-specific recommendations used for Qwen3.5-4B. The context window was 262,144 tokens, with the qwen3 and qwen3_coder parsers.
During evaluation, Think used the remaining context space for output, without an additional cap. Instruct generally used a 32,768-token limit. For MATH-PT, Instruct used the math configuration: temperature 1.0, top-p 1.0, top-k 40, presence penalty 2.0, and an 81,920-token limit. The llama.cpp usage examples below retain the recommended 32,768-token limit.
| Parameter | Think | Instruct |
|---|---|---|
enable_thinking | true (default) | false |
temperature | 1.0 | 0.7 |
top_p | 0.95 | 0.8 |
top_k | 20 | 20 |
min_p | 0.0 | 0.0 |
presence_penalty | 1.5 | 1.5 |
repetition_penalty | 1.0 | 1.0 |
max_tokens | 32,768 | 32,768 |
chat_template_kwargs; the same server
supports both modes.max_tokens can be increased to 81,920.generation_config.json includes the Think-mode values, but not
presence_penalty. Send the parameters with every request.bash
vllm serve maritaca-ai/S5-nano \ --max-model-len 262144 \ --reasoning-parser qwen3 \ --enable-auto-tool-choice --tool-call-parser qwen3_coderThe server starts at http://localhost:8000/v1 and is compatible with the OpenAI API.
Optional flags:
--speculative-config '{"method":"mtp","num_speculative_tokens":3}' enables
speculative decoding with the MTP layer;--language-model-only excludes the vision encoder, freeing memory for
the KV cache;--max-model-len 32768 reduces the context length for GPUs with less memory.For the FP8 version, replace the model name with maritaca-ai/S5-nano-FP8
and keep the same flags.
When serving FP8, also use maritaca-ai/S5-nano-FP8 in the request's
model field (or MODEL in the Python examples).
Think mode:
bash
curl http://localhost:8000/v1/chat/completions \ -H "Content-Type: application/json" \ -d '{ "model": "maritaca-ai/S5-nano", "messages": [ {"role": "user", "content": "Explain why the sky is blue."} ], "temperature": 1.0, "top_p": 0.95, "top_k": 20, "min_p": 0.0, "presence_penalty": 1.5, "max_tokens": 32768 }'Reasoning is returned in choices[0].message.reasoning, and the final answer
in choices[0].message.content.
Instruct mode:
bash
curl http://localhost:8000/v1/chat/completions \ -H "Content-Type: application/json" \ -d '{ "model": "maritaca-ai/S5-nano", "messages": [ {"role": "user", "content": "Write a haiku about the Cerrado."} ], "temperature": 0.7, "top_p": 0.8, "top_k": 20, "min_p": 0.0, "presence_penalty": 1.5, "max_tokens": 32768, "chat_template_kwargs": {"enable_thinking": false} }'top_k, min_p, and chat_template_kwargs are not part of the OpenAI API;
pass them in extra_body.
python
from openai import OpenAI client = OpenAI(base_url="http://localhost:8000/v1", api_key="EMPTY")MODEL = "maritaca-ai/S5-nano" THINK = dict( temperature=1.0, top_p=0.95, presence_penalty=1.5, max_tokens=32768, extra_body={"top_k": 20, "min_p": 0.0},)INSTRUCT = dict( temperature=0.7, top_p=0.8, presence_penalty=1.5, max_tokens=32768, extra_body={ "top_k": 20, "min_p": 0.0, "chat_template_kwargs": {"enable_thinking": False}, },) # Thinkmessages = [{"role": "user", "content": "Explain why the sky is blue."}]resp = client.chat.completions.create(model=MODEL, messages=messages, **THINK)msg = resp.choices[0].messagereasoning = getattr(msg, "reasoning", None)print("reasoning:", reasoning)print("answer:", msg.content) # Next turn: pass the reasoning back along with the answermessages.append({"role": "assistant", "content": msg.content, "reasoning": reasoning})messages.append({"role": "user", "content": "And why does the sunset look reddish?"})resp = client.chat.completions.create(model=MODEL, messages=messages, **THINK)print(resp.choices[0].message.content) # Instructresp = client.chat.completions.create( model=MODEL, messages=[{"role": "user", "content": "Write a haiku about the Cerrado."}], **INSTRUCT,)print(resp.choices[0].message.content)The model was trained with reasoning from previous turns retained in the
conversation history. In multi-turn conversations, include the reasoning
field from each assistant response, as shown above.
The maritaca-ai/S5-nano-GGUF repository provides three quantization variants and the vision projector:
| File | Size | Notes |
|---|---|---|
S5-nano-BF16.gguf | 8.7 GB | lossless |
S5-nano-Q8_0.gguf | 4.6 GB | nearly lossless |
S5-nano-Q4_K_M.gguf | 2.8 GB | smaller, with imatrix |
mmproj-S5-nano-F16.gguf | 0.7 GB | vision encoder |
bash
llama-server -hf maritaca-ai/S5-nano-GGUF:Q4_K_M --jinja -c 32768 -ngl 99Q4_K_M with Q8_0 or BF16 to use a different quantization.-hf also downloads mmproj; use --no-mmproj for text-only serving.--spec-type draft-mtp --spec-draft-n-max 3 enables speculative decoding
with the MTP layer, which is included in all three GGUFs.The server starts at http://localhost:8080/v1. The curl and Python examples
above work after changing the port, with the same parameters and
chat_template_kwargs. The reasoning field has a different name:
llama-server uses reasoning_content instead of reasoning, both in the
response and in the conversation history.
vllm serve maritaca-ai/S5-nano \ --max-model-len 262144 \ --reasoning-parser qwen3 \ --enable-auto-tool-choice --tool-call-parser qwen3_codercurl http://localhost:8000/v1/chat/completions \ -H "Content-Type: application/json" \ -d '{ "model": "maritaca-ai/S5-nano", "messages": [ {"role": "user", "content": "Explain why the sky is blue."} ], "temperature": 1.0, "top_p": 0.95, "top_k": 20, "min_p": 0.0, "presence_penalty": 1.5, "max_tokens": 32768 }'curl http://localhost:8000/v1/chat/completions \ -H "Content-Type: application/json" \ -d '{ "model": "maritaca-ai/S5-nano", "messages": [ {"role": "user", "content": "Write a haiku about the Cerrado."} ], "temperature": 0.7, "top_p": 0.8, "top_k": 20, "min_p": 0.0, "presence_penalty": 1.5, "max_tokens": 32768, "chat_template_kwargs": {"enable_thinking": false} }'from openai import OpenAI client = OpenAI(base_url="http://localhost:8000/v1", api_key="EMPTY")MODEL = "maritaca-ai/S5-nano" THINK = dict( temperature=1.0, top_p=0.95, presence_penalty=1.5, max_tokens=32768, extra_body={"top_k": 20, "min_p": 0.0},)INSTRUCT = dict( temperature=0.7, top_p=0.8, presence_penalty=1.5, max_tokens=32768, extra_body={ "top_k": 20, "min_p": 0.0, "chat_template_kwargs": {"enable_thinking": False}, },) # Thinkmessages = [{"role": "user", "content": "Explain why the sky is blue."}]resp = client.chat.completions.create(model=MODEL, messages=messages, **THINK)msg = resp.choices[0].messagereasoning = getattr(msg, "reasoning", None)print("reasoning:", reasoning)print("answer:", msg.content) # Next turn: pass the reasoning back along with the answermessages.append({"role": "assistant", "content": msg.content, "reasoning": reasoning})messages.append({"role": "user", "content": "And why does the sunset look reddish?"})resp = client.chat.completions.create(model=MODEL, messages=messages, **THINK)print(resp.choices[0].message.content) # Instructresp = client.chat.completions.create( model=MODEL, messages=[{"role": "user", "content": "Write a haiku about the Cerrado."}], **INSTRUCT,)print(resp.choices[0].message.content)llama-server -hf maritaca-ai/S5-nano-GGUF:Q4_K_M --jinja -c 32768 -ngl 99