Nile Chat 12B is an open-source language model by MBZUAI-Paris. Features: 12b LLM, VRAM: 23.4GB, Context: 128K, License: gemma, LLM Explorer Score: 0.21.
Arxiv:2308.16149 Arxiv:2507.04569 Arz Base model:finetune:google/gem... Base model:google/gemma-3-12b-... Conversational Dataset:mbzuai-paris/egyptian-... Endpoints compatible Gemma3 text Region:us Safetensors Sharded Tensorflow
Model Card on HF 🤗: https://huggingface.co/MBZUAI-Paris/Nile-Chat-12B
Nile Chat 12B Benchmarks
nn.n% — How the model compares to the reference models: Anthropic Sonnet 3.5 ("so35"), GPT-4o ("gpt4o") or GPT-4 ("gpt4").
What is the LLM Explorer Rank (Score)
Nile Chat 12B Parameters and Internals
| LLM Name | Nile Chat 12B |
| Repository 🤗 | https://huggingface.co/MBZUAI-Paris/Nile-Chat-12B |
| Base Model(s) | |
| Model Size | 12b |
| Required VRAM | 23.4 GB |
| Updated | 2026-04-01 |
| Maintainer | MBZUAI-Paris |
| Model Type | gemma3_text |
| Model Files | |
| Model Architecture | Gemma3ForCausalLM |
| License | gemma |
| Context Length | 131072 |
| Model Max Length | 131072 |
| Transformers Version | 4.51.0.dev0 |
| Tokenizer Class | GemmaTokenizer |
| Padding Token | <pad> |
| Vocabulary Size | 262208 |
| Torch Data Type | bfloat16 |
Best Alternatives to Nile Chat 12B
Best Alternatives |
Context / RAM |
Downloads |
Likes |
|---|---|---|---|
| GaMS3 12B Instruct | 128K / 23.4 GB | 3410 | 2 |
| YanoljaNEXT Rosetta 12B 2510 | 128K / 25.4 GB | 235 | 29 |
| Typhoon2.1 Gemma3 12B | 128K / 23.4 GB | 1610 | 18 |
| Gemma 3 12B It Abliterated V2 | 128K / 47.2 GB | 890 | 19 |
| Typhoon2.1 Gemma3 12B | 128K / 23.4 GB | 186 | 19 |
| Gemma 3 Coldbrew Test1 | 128K / 23.4 GB | 5 | 0 |
| Gemma 3 Glitter 12B | 128K / 23.4 GB | 139 | 7 |
| Gemma 3 Starshine 12B Alt | 128K / 23.4 GB | 4 | 6 |
| G3 12B Inkstruct Epoch2 | 128K / 23.4 GB | 5 | 0 |
| ... 12B Storyteller V0.2 Textonly | 128K / 23.4 GB | 6 | 6 |
Note: green Score (e.g. "73.2") means that the model is better than MBZUAI-Paris/Nile-Chat-12B .
Rank the Nile Chat 12B Capabilities
🆘 Have you tried this model? Rate its performance. This feedback would greatly assist ML community in identifying the most suitable model for their needs. Your contribution really does make a difference! 🌟
What open-source LLMs or SLMs are you in search of? 52721 in total.
Original data from HuggingFace, OpenCompass and various public git repos.
Check out Ag3ntum — our secure, self-hosted AI agent for server management.
Release v20260328a