HF Model & Benchmark Matrix

24,078 variants · 14,984 scores · 120 benchmarks
Home/Orgs/LiquidAI/LiquidAI/LFM2-1.2B

LiquidAI/LFM2-1.2B

family misc · 12 variants · unknown

provider
LiquidAI
released
—
huggingface
https://huggingface.co/LiquidAI/LFM2-1.2B
best scores
AlpacaEval: 3.76BBH: 49.54IFEval: 98.85WER: 7.53WER AMI: 15.63WER Earnings-22: 14.56WER GigaSpeech: 10.67WER LibriSpeech: 2.01WER SPGISpeech: 2.76WER TED-LIUM: 3.56WER VoxPopuli: 7.13MMLU: 55.23GPQA: 31.47IFBench: 20.7GSM8K: 58.3MGSM: 55.04

Best scores per benchmark

benchmarkcategorybest score
IFEvalinstruction-following98.85
GSM8Kmath58.3
MMLUknowledge55.23
MGSMmath55.04
BBHreasoning49.54
GPQAreasoning31.47
IFBenchinstruction-following20.7
WER AMIspeech-asr15.63
WER Earnings-22speech-asr14.56
WER GigaSpeechspeech-asr10.67
WERspeech-asr7.53
WER VoxPopulispeech-asr7.13
AlpacaEvalchat3.76
WER TED-LIUMspeech-asr3.56
WER SPGISpeechspeech-asr2.76
WER LibriSpeechspeech-asr2.01

Variants (12)

variantkindquantbitsvRAMdlorgyear
LFM2-1.2B-MLX-8bitmlxmlx20,771lmstudio-community2025
LFM2-1.2B-MLX-bf16mlxmlx20,523lmstudio-community2025
LFM2.5-Audio-1.5Boriginalbf1616.01,282LiquidAI2026
LFM2.5-Audio-1.5B-GGUFggufgguf16.0669Mungert2026
LFM2-1.2B-GGUFggufgguf16.0598unsloth2025
LFM2-1.2B-unsloth-bnb-4bitbnb-4bit4bit269unsloth2025
LFM2-Audio-1.5Boriginalbf1616.0253LiquidAI2026
LFM2.5-Audio-1.5B-bf16mlxmlx1.0147mlx-community2026
LFM2-1.2B-Extract-GGUFggufgguf16.0110Mungert2025
LFM2-1.2Boriginalbf1616.098unsloth2025
LFM2-1.2B-Tool-GGUFggufgguf16.061Mungert2025
LFM2-1.2B-RAG-GGUFggufgguf16.021Mungert2025
Hugging Face model & benchmark data scraped & matrix-ified. 24,078 variants · 14,984 scores · 120 benchmarks.