GLM-5.3 is live. Run Z.ai's latest model on Friendli Model APIs. Try it today ➜
Erog0291
Available on FriendliAI
Run this model inference on single tenant GPU with unmatched speed and reliability at scale.
Model Details
Model Provider
Erog0291
Model Tree
Qwen/Qwen3-1.7B
Input Modalities
Output Modalities
Supported Functionality
Ollama (one line, no signup):
text
ollama run emmanuelerogian723/stewmind-1.7bModel page: https://ollama.com/emmanuelerogian723/stewmind-1.7b
Transformers:
python
from transformers import AutoModelForCausalLM, AutoTokenizertok = AutoTokenizer.from_pretrained("Erog0291/stewmind-1.7b-ultra")model = AutoModelForCausalLM.from_pretrained("Erog0291/stewmind-1.7b-ultra")msgs = [{"role": "user", "content": "Route this request: 'Send 5000 naira to my brother today'. Reply with only JSON."}]ids = tok.apply_chat_template(msgs, add_generation_prompt=True, return_tensors="pt")print(tok.decode(model.generate(ids, max_new_tokens=80, do_sample=False)[0][ids.shape[1]:], skip_special_tokens=True))| Test | Result |
|---|---|
| Structured request parsing ("Send 5000 naira...") | Valid JSON, correct fields |
| Tool routing / ReAct planning | Coherent plans |
| Pidgin ("Wetin be your main work?") | Understood and answered |
| Footprint | ~1 GB at Q4_K_M, ~25 tok/s CPU |
| Cost per 1M tokens | $0.00 — runs offline |
| StewMind ULTRA | GPT-6.1 Sol (API) | |
|---|---|---|
| Runs offline / on a phone | YES | No |
| Data leaves your device | Never | Every call |
| Cost per 1M tokens | $0.00 | Paid, rising |
| Native 3D point-cloud reading | Encoders trained (fusion pending) | Cannot ingest 3D at all |
| Nigerian languages (Hausa/Igbo/Yoruba/Pidgin) | Native stack | Generic |
| Deep essay reasoning | Loses, honestly | Wins |
A 1.7B will not out-reason a frontier model. StewMind ULTRA is built for agents: fast, cheap, private, local, with reliable structured output. Deep reasoning is what you call a giant for.
Built with love in Nigeria by MUTYINT. Apache 2.0, commercial use allowed.
from transformers import AutoModelForCausalLM, AutoTokenizertok = AutoTokenizer.from_pretrained("Erog0291/stewmind-1.7b-ultra")model = AutoModelForCausalLM.from_pretrained("Erog0291/stewmind-1.7b-ultra")msgs = [{"role": "user", "content": "Route this request: 'Send 5000 naira to my brother today'. Reply with only JSON."}]ids = tok.apply_chat_template(msgs, add_generation_prompt=True, return_tensors="pt")print(tok.decode(model.generate(ids, max_new_tokens=80, do_sample=False)[0][ids.shape[1]:], skip_special_tokens=True))