SAVRN Model Hub
Open-Weight Model Comparisons
300 pairs of models that do the same job at a similar size, compared on the facts their publishers report: parameters, context length, architecture, license and files, with the memory each needs and the cheapest GPUs to run it at today's prices.
Any to any
- gemma-4-E4B-it vs MiniCPM-o-2_6
- gemma-4-E4B-it vs MiniCPM-o-4_5
- gemma-4-E4B-it vs Qwen2.5-Omni-7B
- gemma-4-E2B-it vs Qwen2.5-Omni-3B
- gemma-4-E2B-it vs gemma-4-E4B
- gemma-4-E2B-it vs gemma-4-E4B-it-uncensored-heretic
- gemma-4-12B-it vs Qwen2.5-Omni-7B
- gemma-4-12B-it vs MiniCPM-o-4_5
- gemma-4-12B-it vs gemma-4-E4B-it-qat-w4a16-ct
Audio classification
- clap-htsat-fused vs wav2vec2-large-robust-12-ft-emotion-msp-dim
- clap-htsat-fused vs Common-Voice-Gender-Detection
- wav2vec2-large-robust-24-ft-age-gender vs wav2vec2-large-xlsr-53-gender-recognition-librispeech
- audiobox-aesthetics vs clap-htsat-fused
- MuQ-large-msd-iter vs wav2vec2-large-robust-24-ft-age-gender
- hubert-large-speech-emotion-recognition-russian-dusha-finetuned vs wav2vec2-large-robust-24-ft-age-gender
Feature extraction
- bge-small-en-v1.5 vs jina-embeddings-v2-small-en
- bge-small-en-v1.5 vs bge-small-zh-v1.5
- bge-small-en-v1.5 vs granite-embedding-small-english-r2
- bge-large-en-v1.5 vs mxbai-embed-large-v1
- bge-large-en-v1.5 vs UAE-Large-V1
- bge-large-en-v1.5 vs multilingual-e5-large-instruct
- bge-base-en-v1.5 vs SapBERT-from-PubMedBERT-fulltext
- bge-base-en-v1.5 vs jina-embeddings-v2-base-code
- bge-base-en-v1.5 vs granite-embedding-small-english-r2
- Qwen3-Embedding-0.6B vs w2v-bert-2.0
- multilingual-e5-large vs w2v-bert-2.0
- granite-embedding-small-english-r2 vs jina-embeddings-v2-small-en
- bge-small-zh-v1.5 vs jina-embeddings-v2-small-en
- bge-reranker-large vs Qwen3-Embedding-0.6B
- bge-reranker-large vs multilingual-e5-large
- bge-reranker-large vs multilingual-e5-large-instruct
- bge-base-en vs SapBERT-from-PubMedBERT-fulltext
- bge-base-en vs jina-embeddings-v2-base-code
- bge-base-en vs mxbai-embed-large-v1
- mxbai-embed-large-v1 vs UAE-Large-V1
- Qwen3-Embedding-4B vs Qwen3-Embedding-8B
- jina-embeddings-v3 vs Qwen3-Embedding-0.6B
- jina-embeddings-v3 vs multilingual-e5-large
- Qwen3-Embedding-4B-W4A16-G128 vs Qwen3-Embedding-8B
Fill mask
- bert-base-uncased vs camembert-base
- distilbert-base-uncased vs distilroberta-base
- distilbert-base-uncased vs SecBERT
- roberta-large vs xlm-roberta-base
- bert-base-multilingual-uncased vs ModernBERT-base
- bert-base-cased vs bert-base-uncased
- bert-base-cased vs bert-base-german-cased
- bert-base-cased vs camembert-base
- bert-base-multilingual-cased vs bert-base-multilingual-uncased
- esm2_t33_650M_UR50D vs xlm-roberta-large
- bert-large-uncased vs xlm-roberta-base
- bert-large-uncased vs roberta-large
- bert-large-uncased vs xlm-roberta-large
- bert-base-chinese vs distilbert-base-uncased
- ModernBERT-large vs xlm-roberta-large
- bert-base-arabertv02 vs roberta-base
- bert-base-arabertv02 vs ModernBERT-base
- bert-base-arabertv02 vs bert-base-multilingual-uncased
- distilbert-base-multilingual-cased vs roberta-base
- distilbert-base-multilingual-cased vs ModernBERT-base
- bert-base-german-cased vs bert-base-uncased
- deberta-v2-large-japanese-char-wwm vs xlm-roberta-base
- deberta-v2-large-japanese-char-wwm vs roberta-large
- codebert-python vs roberta-base
Image and text to text
- gemma-4-26B-A4B-it vs Qwen3.6-35B-A3B-NVFP4
- gemma-4-26B-A4B-it vs Qwen3.8-27B
- gemma-4-26B-A4B-it vs Qwen3.6-27B
- Qwen3.5-9B vs Qwen3-VL-8B-Instruct
- gemma-4-31B-it vs Qwen2.5-VL-32B-Instruct
- gemma-4-31B-it vs Qwen3.6-35B-A3B-NVFP4
- gemma-4-31B-it vs Qwen3.8-27B
- Qwen3.5-4B vs Qwen3-VL-4B-Instruct
- Qwen3.5-4B vs vllm-translategemma-4b-it
- Qwen2.5-VL-7B-Instruct vs Qwen3-VL-8B-Instruct
- Qwen2.5-VL-7B-Instruct vs Qwen3.5-9B
- Qwen2.5-VL-7B-Instruct vs Qwen2-VL-7B-Instruct
- Qwen3.5-2B vs Rax-4.5
- Qwen3.5-2B vs Qwen3-VL-2B-Instruct
- Qwen3.6-27B vs Qwen3.8-27B
- Qwen3-VL-2B-Instruct vs Rax-4.5
- Florence-2-base vs SmolVLM2-500M-Video-Instruct
- Florence-2-base vs surya-ocr-2
- Florence-2-base vs Qwen3.5-0.8B
- chandra-ocr-2 vs vllm-translategemma-4b-it
- chandra-ocr-2 vs gemma-3-4b-it
- chandra-ocr-2 vs medgemma-4b-it
- GLM-5.3-Flash vs Qwen3.8-Flash-Next
- Qwen3.5-0.8B vs surya-ocr-2
- DeepSeek-OCR vs Unlimited-OCR
- Qwen2-VL-2B-Instruct vs Qwen3.5-2B
- Qwen2-VL-2B-Instruct vs Qwen3-VL-2B-Instruct
- Qwen3.5-35B-A3B vs Qwen3.6-35B-A3B
- Qwen3.5-27B vs Qwen3.6-27B
- Qwen3.5-27B vs Qwen3.6-35B-A3B
- gemma-3-4b-it vs Qwen3.5-4B
- gemma-3-4b-it vs Qwen3-VL-4B-Instruct
- Qwen2.5-VL-32B-Instruct vs Qwen3.6-35B-A3B
- medgemma-4b-it vs Qwen3-VL-4B-Instruct
- Qwen2-VL-7B-Instruct vs Qwen3-VL-8B-Instruct
- Qwen2-VL-7B-Instruct vs Qwen3.5-9B
- blip2-opt-2.7b vs Qwen2.5-VL-3B-Instruct
- LocateAnything-3B vs Qwen2.5-VL-3B-Instruct
- Nanonets-OCR2-3B vs Qwen2.5-VL-3B-Instruct
- PaddleOCR-VL-1.6 vs Qwen3.5-0.8B
Image classification
- mobilenetv3_small_100.lamb_in1k vs vit_tiny_patch16_224.augreg_in21k_ft_in1k
- efficientnet_b3.ra2_in1k vs resnet18.a1_in1k
- efficientnet_b3.ra2_in1k vs resnet18.a3_in1k
- efficientnet_b3.ra2_in1k vs repvgg_a0.rvgg_in1k
- vit-base-patch16-224 vs vit_base_patch8_224.augreg2_in21k_ft_in1k
- nsfw_image_detection vs rorshark-vit-base
- gender-classification vs nsfw_image_detection
- vit-base-nsfw-detector vs vit-base-patch16-224
- efficientnet_b0.ra_in1k vs mobilenetv3_small_100.lamb_in1k
- nsfw_image_detector vs vit-base-patch16-224
- nsfw-classifier vs nsfw_image_detection
- edgenext_small.usi_in1k vs mobilenetv3_small_100.lamb_in1k
Image feature extraction
Sentence similarity
- all-MiniLM-L6-v2 vs paraphrase-MiniLM-L6-v2
- all-MiniLM-L6-v2 vs multi-qa-MiniLM-L6-cos-v1
- all-MiniLM-L6-v2 vs rubert-tiny2
- all-mpnet-base-v2 vs paraphrase-multilingual-MiniLM-L12-v2
- all-mpnet-base-v2 vs paraphrase-mpnet-base-v2
- all-mpnet-base-v2 vs multi-qa-mpnet-base-dot-v1
- multilingual-e5-small vs paraphrase-multilingual-MiniLM-L12-v2
- multilingual-e5-small vs nomic-embed-text-v1.5
- multilingual-e5-base vs paraphrase-multilingual-mpnet-base-v2
- bge-base-en-v1.5-course-recommender-v5 vs pubmedbert-base-embeddings
- bge-base-en-v1.5-course-recommender-v5 vs e5-base-v2
- bge-base-en-v1.5-course-recommender-v5 vs e5-base
- all-MiniLM-L12-v2 vs gte-small
- all-MiniLM-L12-v2 vs e5-small-v2
- all-MiniLM-L12-v2 vs snowflake-arctic-embed-s
- all-distilroberta-v1 vs text2vec-base-chinese
- all-distilroberta-v1 vs pubmedbert-base-embeddings
- all-distilroberta-v1 vs e5-base-v2
- embeddinggemma-300m vs paraphrase-multilingual-mpnet-base-v2
- embeddinggemma-300m vs multilingual-e5-base
- embeddinggemma-300m vs gte-multilingual-base
- gte-multilingual-base vs paraphrase-multilingual-mpnet-base-v2
- gte-multilingual-base vs multilingual-e5-base
- distiluse-base-multilingual-cased-v2 vs nomic-embed-text-v1.5
- distiluse-base-multilingual-cased-v2 vs nomic-embed-text-v1
- distiluse-base-multilingual-cased-v1 vs nomic-embed-text-v1.5
- distiluse-base-multilingual-cased-v1 vs nomic-embed-text-v1
- ko-sroberta-multitask vs paraphrase-multilingual-MiniLM-L12-v2
- ko-sroberta-multitask vs multilingual-e5-small
- ko-sroberta-multitask vs nomic-embed-text-v1
Speech recognition
- mms-300m-1130-forced-aligner vs whisper-small
- mms-300m-1130-forced-aligner vs wav2vec2-xls-r-300m-hebrew
- mms-300m-1130-forced-aligner vs wav2vec2-xls-r-300m-ftspeech
- Qwen3-ASR-1.7B vs whisper-large-v3
- Qwen3-ASR-1.7B vs seamless-m4t-v2-large
- Qwen3-ASR-1.7B vs Voxtral-Mini-4B-Realtime-2602
- parakeet-tdt-0.6b-v3 vs whisper-large-v3-turbo
- wav2vec2-xls-r-300m-hebrew vs whisper-small
- nb-wav2vec2-1b-nynorsk vs whisper-large-v3-turbo
- wav2vec2-xls-r-300m-ftspeech vs whisper-small
- parakeet-ctc-1.1b vs whisper-large-v3
- nemotron-3.5-asr-streaming-0.6b vs whisper-large-v3-turbo
- seamless-m4t-v2-large vs whisper-large-v3
Text classification
- bge-reranker-v2-m3 vs roberta-large-mnli
- bge-reranker-v2-m3 vs Prompt-Guard-86M
- bge-reranker-v2-m3 vs xlm-roberta-base-language-detection
- Prompt-Guard-86M vs xlm-roberta-base-language-detection
- bge-reranker-base vs xlm-roberta-base-language-detection
- bge-reranker-base vs roberta-large-mnli
- distilbert-base-uncased-finetuned-sst-2-english vs phishing-email-detection-distilbert_v2.4.1
- distilbert-base-uncased-finetuned-sst-2-english vs distilroberta-finetuned-financial-news-sentiment-analysis
- distilbert-base-uncased-finetuned-sst-2-english vs robertuito-sentiment-analysis
- Bangla-twoclass-Sentiment-Analyzer vs Prompt-Guard-86M
- Bangla-twoclass-Sentiment-Analyzer vs bge-reranker-base
Text generation
- gpt2 vs SmolLM2-135M
- gpt2 vs SmolLM2-135M-Instruct
- gpt2 vs macbert4csc-base-chinese
- Qwen2.5-7B-Instruct vs Qwen2.5-Coder-7B-Instruct
- Qwen2.5-7B-Instruct vs Qwen2.5-Coder-7B
- Qwen2.5-0.5B-Instruct vs Qwen3-0.6B-Base
- OTel-2.0-LLM-31B-IT vs Qwen3-30B-A3B
- Qwen2.5-1.5B-Instruct vs Qwen3-1.7B-Base
- Llama-3.2-1B-Instruct vs Qwen2.5-1.5B-Instruct
- Llama-3.2-1B-Instruct vs TinyLlama-1.1B-Chat-v1.0
- Llama-3.2-1B-Instruct vs OpenELM-1_1B-Instruct
- gpt-oss-20b vs Ornith-1.5-35B-A3B-NVFP4
- Llama-3.1-8B-Instruct vs Qwen3-8B
- Llama-3.1-8B-Instruct vs Meta-Llama-3-8B-Instruct
- gpt-oss-120b vs NVIDIA-Nemotron-3-Super-120B-A12B-BF16
- gpt-oss-120b vs Qwen-72B
- dolphin-2.9.1-yi-1.5-34b vs Qwen3.6-35B-A3B-abliterated-v4
- dolphin-2.9.1-yi-1.5-34b vs Qwen2.5-32B-Instruct
- dolphin-2.9.1-yi-1.5-34b vs Qwen2.5-Coder-32B-Instruct
- DeepSeek-V4-Flash-0731 vs GLM-5.2-NVFP4
- DeepSeek-V4-Flash-0731 vs MiniMax-M2.7
- DeepSeek-V4-Flash-0731 vs DeepSeek-V4-Flash-DSpark
- NVIDIA-Nemotron-3-Nano-4B-BF16 vs Qwen3-4B
- NVIDIA-Nemotron-3-Nano-4B-BF16 vs Qwen2.5-3B-Instruct
- NVIDIA-Nemotron-3-Nano-4B-BF16 vs Qwen3-4B-Instruct-2507
- Kimi-K3-DSpark vs Qwen3-1.7B
- Kimi-K3-DSpark vs Qwen3-1.7B-Base
- Kimi-K3-DSpark vs Llama-3.2-3B-Instruct
- pythia-160m vs vlt5-base-keywords
- pythia-160m vs SmolLM2-135M
- pythia-160m vs SmolLM2-135M-Instruct
- gemma-3-1b-it vs Qwen3-0.6B
- gemma-3-1b-it vs OpenELM-1_1B-Instruct
- gemma-3-1b-it vs TinyLlama-1.1B-Chat-v1.0
- Qwen2.5-14B-Instruct vs Qwen2.5-Coder-14B-Instruct
- Qwen2.5-14B-Instruct vs Qwen3-14B
- Qwen2.5-32B-Instruct vs Qwen3-32B
- GLM-4.7-Flash vs OTel-2.0-LLM-31B-IT
- Llama-3.2-3B-Instruct vs Qwen3-4B
- Llama-3.2-3B-Instruct vs Qwen2.5-3B-Instruct
- Gemma-4-26B-A4B-NVFP4 vs Qwen2.5-14B-Instruct
- Mistral-7B-Instruct-v0.2 vs Qwen2.5-7B-Instruct
- Mistral-7B-Instruct-v0.2 vs Qwen2.5-Coder-7B-Instruct
- Gemma-4-31B-IT-NVFP4 vs gpt-oss-20b
- Qwen3-4B-Base vs Qwen3-4B-Instruct-2507
- Qwen2.5-Coder-32B-Instruct vs Qwen3-32B
- OpenELM-1_1B-Instruct vs Qwen3-0.6B
- PowerMoE-3b vs Qwen3-4B
- PowerMoE-3b vs Qwen2.5-3B-Instruct
- gpt2-large vs Qwen3-0.6B
- gpt2-large vs Qwen2.5-0.5B-Instruct
- Meta-Llama-3-8B-Instruct vs Qwen3-8B
- NVIDIA-Nemotron-3.5-Lightning-30B-A3B-NVFP4 vs Qwen3.6-35B-A3B-NVFP4
- Ornith-1.5-35B-A3B-NVFP4 vs Qwen3.6-35B-A3B-NVFP4
- DeepSeek-V4-Flash-DSpark vs gpt-oss-120b
- Llama-3.3-70B-Instruct vs Qwen-72B
- qwen3-4b-base-dapo-v4 vs Qwen3-4B-Instruct-2507
- DeepSeek-R1-0528-Qwen3-8B vs Qwen3-8B
- DeepSeek-R1-0528-Qwen3-8B vs Llama-3.1-8B-Instruct
- Qwen2.5-1.5B vs Qwen3-1.7B
- OLMo-2-0425-1B vs Qwen2.5-1.5B-Instruct
- OLMo-2-0425-1B vs Qwen3-1.7B
- Qwen2-0.5B vs Qwen2.5-0.5B-Instruct
- NVIDIA-Nemotron-3-Nano-30B-A3B-NVFP4 vs Qwen3.6-35B-A3B-NVFP4
- NVIDIA-Nemotron-3-Super-120B-A12B-NVFP4 vs Qwen-72B
- gpt-neox-20b vs gpt-oss-20b
- Qwen2.5-7B vs Qwen2.5-Coder-7B-Instruct
- NVIDIA-Nemotron-3-Nano-30B-A3B-BF16 vs OTel-2.0-LLM-31B-IT
- NVIDIA-Nemotron-3-Nano-30B-A3B-BF16 vs Qwen3-32B
Text ranking
- ms-marco-MiniLM-L4-v2 vs ms-marco-MiniLM-L6-v2
- gte-reranker-modernbert-base vs mmarco-mMiniLMv2-L12-H384-v1
- gte-reranker-modernbert-base vs ms-marco-electra-base
- Qwen3-Reranker-4B vs Qwen3-VL-Reranker-2B
- mmarco-mMiniLMv2-L12-H384-v1 vs ms-marco-electra-base
- mmarco-mMiniLMv2-L12-H384-v1 vs qnli-electra-base
- ms-marco-MiniLM-L12-v2 vs ms-marco-MiniLM-L6-v2
- ms-marco-MiniLM-L12-v2 vs ms-marco-MiniLM-L4-v2
- llama-nemotron-rerank-1b-v2 vs Qwen3-Reranker-4B
- ms-marco-MiniLM-L2-v2 vs ms-marco-MiniLM-L6-v2
- ms-marco-MiniLM-L2-v2 vs ms-marco-MiniLM-L4-v2
- cross-encoder-russian-msmarco vs gte-reranker-modernbert-base
Text to image
Text to speech
Time series forecasting
- chronos-2 vs Kronos-base
- chronos-2 vs Toto-Open-Base-1.0
- chronos-2 vs chronos-t5-base
- chronos-2 vs chronos-2-synth
- chronos-2 vs Kronos-base
- chronos-2 vs Toto-Open-Base-1.0
- chronos-bolt-small vs chronos-t5-small
- chronos-bolt-small vs Kronos-small
- chronos-2-small vs chronos-bolt-small
- chronos-2-small vs Kronos-small
- chronos-2-small vs Toto-2.0-22m
- chronos-bolt-small vs chronos-t5-small
- chronos-bolt-small vs Kronos-small
- chronos-bolt-small vs Kronos-base
- Kronos-Tokenizer-base vs Toto-2.0-4m
- Kronos-Tokenizer-2k vs Kronos-Tokenizer-base
- Kronos-mini vs Kronos-Tokenizer-base
Token classification
Translation
Zero shot image classification
- clip-vit-large-patch14 vs siglip2-base-patch16-256
- clip-vit-large-patch14 vs siglip2-base-patch16-naflex
- CLIP-ViT-L-14-laion2B-s32B-b82K vs clip-vit-large-patch14
- CLIP-ViT-L-14-laion2B-s32B-b82K vs siglip2-base-patch16-256
- CLIP-ViT-L-14-laion2B-s32B-b82K vs siglip2-base-patch16-naflex
- siglip2-base-patch16-256 vs siglip2-base-patch16-naflex
- CLIP-ViT-B-32-laion2B-s34B-b79K vs fashion-clip
- CLIP-ViT-B-32-laion2B-s34B-b79K vs siglip-base-patch16-224
- CLIP-ViT-B-32-laion2B-s34B-b79K vs siglip2-base-patch16-224
- siglip2-giant-opt-patch16-384 vs siglip2-so400m-patch14-384
- siglip-so400m-patch14-384 vs siglip2-giant-opt-patch16-384