README.md

metadata

language:
  - tr
  - en
  - de
  - es
  - fr
  - ru
  - zh
  - ja
  - ko
license: mit
tags:
  - turkish
  - türkiye
  - reasoning
  - ai
  - lamapi
  - gemma3
  - next
  - next-x1
  - text-generation
  - open-source
  - 14b
  - large-language-model
  - llm
  - transformer
  - artificial-intelligence
  - machine-learning
  - nlp
  - multilingual
  - instruction-tuned
  - chat
  - generative-ai
  - optimized
  - trl
  - sft
  - cognitive
  - analytical
  - enterprise
pipeline_tag: text-generation
datasets:
  - mlabonne/FineTome-100k
  - CognitiveKernel/CognitiveKernel-Pro-SFT
  - OpenSPG/KAG-Thinker-training-dataset
  - Gryphe/ChatGPT-4o-Writing-Prompts
  - QuixiAI/dolphin-r1
  - uclanlp/Brief-Pro
library_name: transformers

🧠 Next 14B (l310)

Türkiye’s First Reasoning-Capable AI Model — Logical, Analytical, and Enterprise-Ready

---

📖 Overview

Next 14B is a 14-billion parameter large language model (LLM) built upon Qwen 3 architecture, trained to achieve superior reasoning and analytical capabilities.
It is Türkiye’s first reasoning-capable AI model, designed to think, infer, and make decisions — not just respond.

Unlike vision-based models, Next 14B focuses on pure cognitive performance, mastering complex problem solving, abstract logic, and human-level understanding in both Turkish and English.

---

⚡ Highlights

• 🇹🇷 Türkiye’s first reasoning-capable AI model
• 🧠 Advanced logical, analytical, and inferential reasoning
• 🌍 High multilingual understanding (Turkish, English, and beyond)
• 🏢 Enterprise-grade stability and consistency
• 💬 Instruction-tuned for dialogue, problem solving, and analysis

---

📊 Benchmark Performance

Model	MMLU (5-shot) %	MMLU-Pro %	GSM8K %	MATH %
Next 14B (Thinking)	94.6	93.2	98.8	92.7
Next 12B	92.7	84.4	95.3	87.2
Next 8B (Thinking)	91.0	88.5	96.2	88.0
GPT-5	92.5	87.0	98.4	96.0
Claude Opus 4.1 (Thinking)	~92.0	87.8	84.7	95.4

---

🚀 Installation & Usage

from transformers import AutoTokenizer, AutoModelForCausalLM
import torch

model_id = "Lamapi/next-14b"

tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(model_id, torch_dtype=torch.float16, device_map="auto")

messages = [
    {"role": "system", "content": "You are Next-X1, a reasoning-capable AI assistant created by Lamapi. You think deeply, reason logically, and always answer concisely. Proudly made in Turkey."},
    {"role": "user", "content": "Explain why the sky appears blue using logical reasoning."}
]

prompt = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
inputs = tokenizer(prompt, return_tensors="pt").to(model.device)

outputs = model.generate(**inputs, max_new_tokens=150)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))

---

🧩 Key Features

Feature	Description
🧠 Advanced Reasoning	Excels in abstract logic, critical thinking, and long-form analysis.
🇹🇷 Cultural & Multilingual Intelligence	Deep Turkish understanding, alongside fluent English and 30+ languages.
⚙️ Optimized for Efficiency	Available in quantized formats (Q8_0, Q4_K_M, FP16).
🧮 Mathematical & Analytical Skill	Performs exceptionally in structured problem solving and scientific reasoning.
🧩 Non-Vision Architecture	Focused purely on cognitive and linguistic understanding.
🏢 Enterprise Reliability	Consistent, interpretable outputs for professional use cases.

---

📐 Model Specifications

Specification	Details
Base Model	Qwen 3
Parameters	14 Billion
Architecture	Transformer (Causal LLM)
Modalities	Text-only
Fine-Tuning	Instruction-tuned and reinforced with cognitive reasoning datasets
Optimizations	Quantization-ready, FP16 support
Primary Focus	Reasoning, logic, decision-making, and language understanding

---

🎯 Ideal Use Cases

• Analytical Chatbots for business and enterprise logic
• Research Assistance — scientific, legal, or data-heavy reasoning
• Education & Tutoring — explain concepts step-by-step
• Creative Writing — coherent story logic and worldbuilding
• Code & Algorithm Design — reasoning-based code generation
• Decision Support Systems — scenario evaluation and inference

---

💡 Performance Highlights

• Superior Reasoning: Outperforms previous-generation 12B models in logic-based benchmarks.
• Robust Mathematical Understanding: Handles symbolic reasoning and complex equations.
• Consistent Long-Context Memory: Capable of tracking context across multi-turn conversations.
• Professional Reliability: Built for critical enterprise and research applications.

---

📄 License

Licensed under the MIT License — free for commercial and non-commercial use. Attribution is appreciated.

---

📞 Contact & Support

• 📧 Email: lamapicontact@gmail.com
• 🤗 HuggingFace: Lamapi

---

Next 14B — Türkiye’s first reasoning-capable large language model, combining logical depth, analytical intelligence, and enterprise reliability.