Foundation Models
Free
Gemma 3
Models (7)
- Gemma 3 27BchatElo 1358
- Gemma 3 12BchatElo 1334
- Gemma 3n 4BchatElo 1306
- Gemma 3 4BchatElo 1291
- Gemma 2 27BchatElo 1231
- Gemma 4 26B A4B (free)chat
- Gemma 4 31B (free)chat
Why is this hot
32 mentions / 7d across 3 sources (diversity ×1.3)last 48h: 9 vs baseline 3.3/day
- Show HN: Gemma 4 26B at 100 tok/s on a Mac, using the Neural Engine and 2GB RAMhn
- Show HN: Maali – Privacy-first personal finance app with a local AI analysthn
- I tried hacking together a basic React component, but ChatGPT hit a refusal loopbluesky
- Task-Preserving Multilingual Instruction Tuning for Open-Ended QA in WMT 2026 MIarxiv
- Benchmarking Large Language Models for Detecting Satirical Code-Mixed Sentencesarxiv
Top mentions by weighted contribution (source authority × engagement × recency).
Description
Google's open-weight model family.
Recent mentions in news (15)
- Polar Switch: Fast Muon Orthogonalization with Polynomial Circuitsarxiv-ai · 2d
- HydraLLM: Adaptive Width Slicing for Elastic Text and Vision-Language Modelsarxiv-ai · 5d
- Taking Tokenization off the Host for Agentic LLM Servingarxiv-ai · 12d
- Show HN: Breadcrumb – A flight recorder and context manager for your AI (Mac)hn-ai · 20d
- Permutation-Robust Decision Modeling with Candidate-Independent Block-Causal Attentionarxiv-ai · 20d
- LampAttention: Look-Ahead Mixed-Precision FlashAttention for Dedicated Acceleratorsarxiv-ai · 21d
- The Invisible Language Tax: Token Premiums of French and Regional Languages in 2026 LLM Tokenizers, and a French-Optimized Prototypearxiv-ai · 22d
- Benchmarking and Domain Adaptation of Automatic Speech Recognition (ASR) for Adolescent Health Communication in Ghanaian Languagesarxiv-ai · 27d
- Hallucination Neurons and Where to Find Them: An Investigation into the existence of Hallucination Neuronsarxiv-ai · 27d
- Repurposing Pre-trained LLMs as High Fidelity Continuous Text Autoencodersarxiv-ai · 29d
- Show HN: Gemma 3 4B as a typed decision function in Rust (47 ms/decision)hn-ai · 29d
- Disaggregated Quantization: Specializing LLM Prefill and Decodearxiv-ai · 29d
- Modular Norm RandOpt: Population-Efficient Ensembling through Architecture-Aware Perturbationsarxiv-ai · 1mo
- When Better Turns Do Not Make Better Agents: Diagnosing the Gap Between Next-Turn Metrics and Workflow Successarxiv-ai · 1mo
- Open-jev: One-pass option scoring with Gemma 3 4B, similar to jevhn-ai · 1mo
Related Briefs & Stories (6)
7.3
1 items • 3moinbox
6.3
1 items • 3moinbox
4.8
1 items • 3moUnbriefed story cluster
4.5
1 items • 2moUnbriefed story cluster
4.3
1 items • 2moUnbriefed story cluster