Meta tags:
description=A curated reference library for language models — open source, proprietary, local, and cloud. Compare specs, benchmarks, and use cases.;

Headings (most frequently used words):

model, every, that, matters, one, place, benchmarks, run, locally, with, ollama, contact, quick, start, hardware, guide, got, to, suggest, found, an, error, send, message,

Text of the page (most frequently used words):
model (13), ollama (9), pull (6), vram (5), models (5), and (4), run (4), open (4), source (4), #locally (4), updated (3), data (3), from (3), benchmarks (3), erlib (3), message (3), library (3), the (3), proprietary (3), llm (3), published (2), cards (2), com (2), send (2), suggest (2), field (2), moves (2), you (2), gpus (2), llama3 (2), with (2), mmlu (2), language (2), humaneval (2), coding (2), context (2), type (2), 2025, community, protected, cloudflare, turnstile, tracking, cookies, other, report, incorrect, topic, email, your, name, this, maintained, manually, know, worth, indexing, spotted, outdated, just, want, say, hello, reach, out, got, found, error, contact, 13b, 30b, 70b, split, across, required, depends, size, most, consumer, handle, 14b, comfortably, cpu, inference, slower, but, works, hardware, guide, deepseek, mistral, gemma3, qwen3, lets, single, command, install, then, any, compatible, quick, start, massive, multitask, understanding, accuracy, pass, scores, sourced, third, party, evals, may, vary, methodology, provider, multimodal, reasoning, runs, feature, all, curated, index, compare, windows, use, cases, every, that, matters, one, place, reference, indexed,

Text of the page (random words):
erlib the llm library erlib llm 16 models indexed llm reference library every model that matters one place a curated index of language models open source and proprietary compare context windows use cases and benchmarks updated as the field moves 16 models 9 open source 7 proprietary 9 run locally type all open source proprietary feature runs locally reasoning coding multimodal benchmarks model provider type mmlu humaneval context mmlu massive multitask language understanding accuracy humaneval coding pass 1 scores sourced from published model cards and third party evals may vary by methodology run locally with ollama quick start ollama lets you run open source models locally with a single command install from ollama com then pull any compatible model ollama pull llama3 3 ollama pull deepseek r1 ollama pull mistral ollama pull gemma3 ollama pull qwen3 ollama run llama3 3 hardware guide required vram depends on model size most consumer gpus handle 7 14b models comfortably cpu inference is slower but works 7b model 5 gb vram 13b model 9 gb vram 30b model 20 gb vram 70b model 40 gb vram or split across gpus contact got a model to suggest found an error this library is maintained manually and updated as the field moves if you know a model worth indexing spotted outdated data or just want to say hello reach out send a message your name email topic suggest a model report incorrect data other message send message protected by cloudflare turnstile no tracking no cookies erlib com updated 2025 data from published model cards community benchmarks