1,226 models · 1,226 open weights · 200 API-only · refreshed nightly
Open-weights models
Models whose weights are published on Hugging Face — download them and run them on your own hardware. Many are also served by hosted providers; those carry an "Also via API" badge with live prices.
| # | Model | Params | Context | Commercial use | 30d | Min VRAM |
|---|---|---|---|---|---|---|
| 01 | bge-reranker-v2-m3 | 570 M | 8 K | ✓ apache-2.0 | 14.7 M | from 3.1 GB |
| 02 | roberta-base-go_emotions | 120 M | 512 | ✓ mit | 4.6 M | from 1.1 GB |
| 03 | finbert | 110 M | 512 | ✗ cc-by-nc-sa-4.0 | 3.9 M | from 1.0 GB |
| 04 | bge-reranker-base | 280 M | 512 | ✓ mit | 3.0 M | from 1.8 GB |
| 05 | distilbert-base-uncased-finetuned-sst-2-english | 70 M | 512 | ✓ apache-2.0 | 2.6 M | from 0.8 GB |
| 06 | Prompt-Guard-86M | 280 M | — | ⚠ llama3.1 | 2.4 M | from 1.8 GB |
| 07 | twitter-roberta-base-sentiment-latest | — | 512 | ✓ cc-by-4.0 | 2.4 M | — |
| 08 | Bangla-twoclass-Sentiment-Analyzer | 280 M | 512 | ✓ mit | 1.9 M | from 1.8 GB |
| 09 | turn-detector | 130 M | 8 K | other | 1.2 M | from 1.1 GB |
| 10 | GEV-26B-Decide | 25.8 B | — | ✓ apache-2.0 | 1.2 M | from 61.4 GB |
| 11 | unbiased-toxic-roberta | — | 512 | ✓ apache-2.0 | 871 K | — |
| 12 | JEV-35B | 36.0 B | — | ✓ apache-2.0 | 776 K | from 85.0 GB |
| 13 | Qwen2.5-Math-PRM-7B | 7.6 B | 4 K | other | 719 K | from 18.4 GB |
| 14 | ms-marco-MiniLM-L-6-v2 | — | 512 | unknown | 710 K | — |
| 15 | finbert-tone | — | 512 | unknown | 702 K | — |
| 16 | bert-base-multilingual-uncased-sentiment | 170 M | 512 | ✓ mit | 673 K | from 1.3 GB |
| 17 | deberta-v3-base-prompt-injection-v2 | 180 M | 512 | ✓ apache-2.0 | 650 K | from 1.3 GB |
| 18 | twitter-xlm-roberta-base-sentiment | — | 512 | unknown | 637 K | — |
| 19 | distilbert-base-multilingual-cased-sentiments-student | 140 M | 512 | ✓ apache-2.0 | 565 K | from 1.1 GB |
| 20 | JEV-9B | 9.0 B | 262 K | ✓ apache-2.0 | 564 K | from 22.1 GB |
| 21 | emotion-english-distilroberta-base | — | 512 | unknown | 503 K | — |
| 22 | fasttext-language-identification | — | — | ✗ cc-by-nc-4.0 | 500 K | — |
| 23 | MedCPT-Cross-Encoder | — | 512 | other | 491 K | — |
| 24 | xlm-emo-t | — | 512 | unknown | 458 K | — |
| 25 | german-sentiment-bert | 109 M | 512 | ✓ mit | 431 K | from 1.0 GB |
| 26 | glotlid | — | — | other | 416 K | — |
| 27 | robertuito-sentiment-analysis | 110 M | 128 | unknown | 392 K | from 1.0 GB |
| 28 | roberta-base-openai-detector | 125 M | 512 | ✓ mit | 381 K | from 1.1 GB |
| 29 | cryptobert | 120 M | 512 | ✓ mit | 376 K | from 1.1 GB |
| 30 | inclusively-classification | 110 M | 512 | ✗ cc-by-nc-sa-4.0 | 348 K | from 1.0 GB |
| 31 | deberta-large-mnli | — | 512 | ✓ mit | 340 K | — |
| 32 | JEV-27B | 26.9 B | 262 K | ✓ apache-2.0 | 340 K | from 64.7 GB |
| 33 | koelectra-polite-v1 | — | 512 | unknown | 330 K | — |
| 34 | prokbert-mini-promoter | 20 M | 1 K | ✗ cc-by-nc-4.0 | 302 K | from 0.6 GB |
| 35 | phishing-email-detection-distilbert_v2.4.1 | 70 M | 512 | ✓ apache-2.0 | 287 K | from 0.8 GB |
| 36 | koelectra-small-v3-nsmc | 10 M | 512 | ✓ mit | 283 K | from 0.6 GB |
| 37 | phishing-distilbert-cyber207 | 67 M | 512 | unknown | 274 K | from 0.8 GB |
| 38 | bertweet-base-sentiment-analysis | — | 128 | unknown | 267 K | — |
| 39 | rubert-tiny-toxicity | 10 M | 512 | ✓ mit | 258 K | from 0.6 GB |
| 40 | FinancialBERT-Sentiment-Analysis | — | 512 | unknown | 253 K | — |
| 41 | Qwen2.5-1.5B-apeach | 1.5 B | 131 K | unknown | 251 K | from 7.5 GB |
| 42 | roberta-large-mnli | 360 M | 512 | ✓ mit | 246 K | from 2.1 GB |
| 43 | mmarco-mMiniLMv2-L6-H384-v1 | — | 512 | unknown | 244 K | — |
| 44 | distilbert-multilingual-toxicity-classifier-ONNX | — | 512 | ⚠ openrail++ | 224 K | — |
| 45 | twitter-roberta-base-sentiment | — | 512 | unknown | 220 K | — |
| 46 | toxic-bert | 110 M | 512 | ✓ apache-2.0 | 216 K | from 1.0 GB |
| 47 | twitter-xlm-roberta-base-sentiment-multilingual | — | 512 | unknown | 215 K | — |
| 48 | xlm-roberta-base-language-detection | 280 M | 512 | ✓ mit | 213 K | from 1.8 GB |
| 49 | decider-2b | 1.9 B | 262 K | ✓ apache-2.0 | 210 K | from 5.7 GB |
| 50 | ftrobertallm | 355 M | 512 | ✓ mit | 202 K | from 2.1 GB |
VRAM figures are estimates for the smallest available quantization at 8K context — see /methodology. Downloads refresh nightly from the Hugging Face API; API prices (blended 3:1 input:output per 1M tokens) and Artificial Analysis indices from OpenRouter.