GLM-5.3 is live. Run Z.ai's latest model on Friendli Model APIs. Try it today ➜
tanishbooti
Available on FriendliAI
Run this model inference on single tenant GPU with unmatched speed and reliability at scale.
Model Details
Model Provider
tanishbooti
Model Tree
Qwen/Qwen2.5-1.5B-Instruct
Input Modalities
Output Modalities
Supported Functionality
This model is not a knowledge source. In the assistant it does two narrow jobs, always on top of live data fetched per query (Open-Meteo weather and air quality, OpenStreetMap geometry, OSRM routing):
{"activity","location","radius_km"} JSON. Output is validated: a location must appear verbatim in the user's text.[E3]) and referring to venues only by codes (P1, P2). A verifier rejects any narration containing a number, place or evidence id that is not in the evidence pack and falls back to a template.Ranking, thresholds and safety vetoes are not decided by the model.
python
from transformers import AutoModelForCausalLM, AutoTokenizertok = AutoTokenizer.from_pretrained("THIS_REPO"); model = AutoModelForCausalLM.from_pretrained("THIS_REPO")Code, scoring method and data sources: see the project repository linked in the submission.