#small-language-model

Local AI should not replace your agent

Local models are usually presented as a destination: install a runtime, open a new chat, choose a model, and move your work there. That can…

Can a 100M Parameter On-Device ASR Model Even Transcribe Rap?

Stop Renting Intelligence You Don’t Need

I’ve watched close to a dozen teams reach for the same default when a new AI feature comes up: pick the biggest, most capable frontier…

The Small Model Revolution: Why Bigger Stopped Being Better in 2026

For three years the answer to “which model?” was “the biggest one you can afford.” In 2026 that answer quietly flipped — and it changes how…

Entropy-Guided Self-Correction: A Lightweight Thinking Mode for Fine-Tuned Small Language Models

I’m a writer. I use AI to brainstorm — not to write for me, but to think alongside me in my own voice. The problem is most AI models don’t…

Not Every Telecom AI Needs a 600B Parameter Model

What if the future of telecom AI isn't about building bigger models—but building smarter systems?

I Wouldn’t Let My Kid Chat With Kimi K3.

Why 2.8 trillion open-weight parameters are raw material, not a product

The CPU Tier for Agents and Small Language Models

Which Agents and Which Small Models Actually Belong There

You Don’t Always Need a Frontier Model. You Need One That Fits.

Building real applications with right-sized, open-weight models using Gemma 4

Intelligence at the Edge

Bringing AI Closer to Where Decisions Happen

The Silicon Tipping Point: How Small Language Models Brought AI On-Device in 2025–2026

Executive Summary

Domain In, IAB Out: Introducing the ZeroGPU Domain IAB Classifier

A specialized small language model that classifies raw domains into IAB categories with higher accuracy than GPT-5.4 Nano, at about 50x the…

The Data Center Is Shrinking Into Your Browser Tab

Introduction: The Death of the Server Rack?

Agentic RAG in Edge Applications

On-Device LLMs, RAG, and the Tools to Make It All Happen

Introducing the ZeroGPU MCP Server

How to connect to our new MCP server in Claude to reduce inference costs and latency for your high-volume AI tasks

Zeki Olmayan Ama Çok Çalışan SLMler

Sadece 3 milyar parametreye sahip Nanbeige ve VibeThinker modellerinin, trilyonluk devlere kodlama ve matematikte nasıl rakip olduklarını…

ZeroGPU Launches Specialized AI Models for Adtech, Cutting Token Costs by More Than 5x

ZeroGPU is building small language models that run at the edge for media, adtech, and other high-volume AI workflows, handling IAB…

Weibo AI — 3B 파라미터로 증명한 소형 모델의 추론 한계 돌파: VibeThinker-3B

1. 한눈에 보는 핵심 요약 (Executive Summary)

Don’t Waste Tokens on Data Entry: Tag Customer Reviews Overnight with ZeroGPU Batch API

A new cookbook to demonstrate our new Batch API.

Fine-Tuning Llama 3.1

There is a moment in every machine learning project where the abstraction disappears.

I distilled a 20B model into 0.5B so my résumé could sing

Apply like the main character: a résumé-to-music-video machine, all under 32B params. A weekend build for the Build Small Hackathon.

Part 3: Fine-Tuning Gemma 4 on a Mac Mini, and What the Results Taught Me

Series: Fine-Tuning a Small Language Model to Sound Like Me

Part 1: Why Prompt Engineering Wasn’t Enough

Series: Fine-Tuning a Small Language Model to Sound Like Me

Fine-Tuning & Evaluation of Small Language Models for Customer Support

Running Language Models Directly in the Browser

TLDR

Why the “Jumbo Jet” Era of AI is Ending: 6 Surprising Takeaways on Small Language Models for 2026

For the last three years, the AI arms race has been defined by “jumbo jets” massive, trillion-parameter cloud models like GPT-4 or Claude…

Why Smaller Language Models Make You Choose Between Accuracy and Explanation Quality

Large language models are impressive, but they are also enormous. Running a model with hundreds of billions of parameters takes serious…

Introducing Batch Processing for ZeroGPU

Running AI inference one request at a time works well for real-time product experiences. But many workloads do not need an immediate…

From a CSV on Your Laptop to a Served Model in 90 Seconds

The hard problem in fine-tuning isn’t model.fit(). It's everything around it: shape your data, build a gold set, pick the right task…

The Watch List: SLMs, Reasoning Models, and the Frontiers Worth Your Attention

Series: The Practical AI Skills Roadmap 2026 | Part 6 of 6

How to Fine-tune a Language Model on a 3 GB GPU

A complete, battle-tested guide to QLoRA fine-tuning using HuggingFace, PEFT, and bitsandbytes

Domain-Specific Small Language Models (SLMs) in Python: Fine-Tuning Phi-3 and Gemma for Industry…

The AI landscape is shifting fast. For years, the dominant narrative was “bigger is better,” with organizations racing to deploy massive…

ZeroGPU Adds Liquid AI Models on its Global Edge Inference Network

Developers can now run Liquid’s LFM2.5 Instruct and Thinking models through ZeroGPU’s global edge inference network, bringing faster…

Enterprise AI Has Entered Its Microservices Moment

Workflow Engines Are Becoming AI Agent Platforms — And SLMs Will Power the Shift

Your Product Doesn’t Need GPT-5. And It’s Costing You More Than You Think.

How Fine-Tuned Small Models Outperform Frontier AI for Most Production Workloads

Turning Two DGX Sparks into a Local LLM Cluster with vLLM, Ray and Qwen3.6

I have been exploring what it takes to move beyond running a local model on a single machine and instead treat multiple small AI systems as…

Teaching Small Models to Do Big Things

How to evaluate Small Language Models using the RAGAS framework — without burning the planet (or your budget) in the process.

Beginner’s Guide to Running Small Language Models Locally

If you’ve heard about AI models like ChatGPT but recently came across ‘Small Language Models’, here’s what that actually means.

The Future of Automation is Local: Why German Firms are Trading the Cloud for On-Premise AI

In early 2026, the AI landscape reached a crossroads. On one side, we have the “reasoning giants”: GPT-5.4 and Gemini 3.1 Pro. These models…

The 4GB Champion: Why Your Next AI Strategy Lives Under a Hard Memory Ceiling

The era of the “cloud-only” AI is ending. For years, the industry has operated on the assumption that intelligence requires scale…

Your LLM Bill is Crushing You. How to Deploy Production AI Without Going Broke.

Moving Beyond the Hype: Mastering Cost-Effective AI Inference with SLMs & Quantization for Real-World Impact.

I Ended Up Writing an On-Device Inference Engine from Scratch

small-infer is a minimal on-device LLM inference engine written in Rust, targeting iOS and Android. Around 4,500 lines of code covering…

Dr. Qwen: Fine-tuning & Evaluating Medical LLMs from 0.6B to 8B with Unsloth

In this project, I’ve fine-tuned three Qwen3 models on a high-quality synthetic medical dataset from OpenMed and had GPT-5.2 judge the…

[Meta] Scaling Small Agents Through Strategy Auctions: SALE 프레임워크의 탄생

Background and Overview

Optimizing Small Language Models for Edge Deployment: A Pipeline for Fine-Tuning and Quantizing…

A technical walkthrough on injecting domain-specific knowledge into 1B-parameter models using Unsloth, LoRA, and GGUF quantization for CPU…

Small Language Models for Enterprise AI Transform Business Operations

Small language models for enterprise AI has an objective of carrying out more important tasks more quickly and successfully, rather than…

From Giants to Gems: Why Small AI Models Are Changing the Game

It is easy to think that progress in artificial intelligence is measured by size. For years, the hype revolved around trillion-parameter…

Language Modeling from the Ground Up

Text is made up of words. When we hear a word, we don’t think of the sound that represents it but instead, what the word itself represents…

SLMs — The Hidden AI Already in Your Devices

If the giants are the headline, the SLMs are the story. And they’re already writing the next chapter of computing.

SLMs als Geheimwaffe: Wie die Mini-KI Anomalien schneller erkennt

TL;DR: SLMs können lokal ausgeführt werden und eignen sich als Erweiterung von Cybersecurity Tools. Dabei können vor allem Anomalien aller…