MDL
gemma-3-270m
Verified against Artificial Analysis · Jul 13, 2026
Parameters
270M
Downloads
2.9M
on HF
Likes
1K
Avg score
5.6
10 benchmarks
AA Intelligence
2.4
index
Weights
Open
Released
Aug 5, 2025
About
gemma-3-270m is a 270M-parameter open-weights model in the gemma family from google. Benchmarked on 10 evals, averaging 5.6.
gemmatext-generation
Benchmark scores · 10
avg 5.6 — higher is betterThe research behind this model
all papers mentioning it →Gemma 3 Technical ReportHellaSwag: Can a Machine Really Finish Your Sentence?BoolQ: Exploring the Surprising Difficulty of Natural Yes/No QuestionsPIQA: Reasoning about Physical Commonsense in Natural LanguageTriviaQA: A Large Scale Distantly Supervised Challenge Dataset for Reading ComprehensionOn the Measure of IntelligenceWinoGrande: An Adversarial Winograd Schema Challenge at ScaleInstruction-Following Evaluation for Large Language ModelsGPQA: A Graduate-Level Google-Proof Q&A BenchmarkMeasuring short-form factuality in large language modelsSocialIQA: Commonsense Reasoning about Social InteractionsDROP: A Reading Comprehension Benchmark Requiring Discrete Reasoning Over ParagraphsMeasuring Massive Multitask Language UnderstandingAGIEval: A Human-Centric Benchmark for Evaluating Foundation ModelsMeasuring Mathematical Problem Solving With the MATH DatasetTraining Verifiers to Solve Math Word ProblemsProgram Synthesis with Large Language ModelsEvaluating Large Language Models Trained on CodeLiveCodeBench: Holistic and Contamination Free Evaluation of Large Language Models for CodeCan LLM Already Serve as A Database Interface? A BIg Bench for Large-Scale Database Grounded Text-to-SQLsNaturalCodeBench: Examining Coding Performance Mismatch on HumanEval and Natural User PromptsLanguage Models are Multilingual Chain-of-Thought ReasonersThe FLORES-101 Evaluation Benchmark for Low-Resource and Multilingual Machine TranslationOn the Cross-lingual Transferability of Monolingual RepresentationsWMT24++: Expanding the Language Coverage of WMT24 to 55 Languages & DialectsChronicle: A Multimodal Foundation Model for Joint Language and Time Series UnderstandingMay 18, 2026
Mentions matched by name in title/abstract — from the arXiv + HF daily corpus.
History
full ledger →Capability over time
Append-only ledger — every observed change to this model's numbers.
API providers
Artificial Analysis$0/M