| 1 | EVIE-4.5Btencent | Visual Document Retrieval | 2 | ▲70.5%(up) | 40 | 4.5B |
| 2 | Soprano-1.1-80Mekwek | Text to Speech | 2 | ▲1.4%(up) | 222 | 79.7M |
| 3 | K2-Horizon-MoVA-36B-A4B-GGUFIFM | Text Generation | 2 | ▲24.0%(up) | 150 | — |
| 4 | NVIDIA-Nemotron-3-Super-120B-A12B-FP8nvidia | Text Generation | 2 | ▲0.5%(up) | 292 | 124B |
| 5 | Fara1.5-27Bmicrosoft | Image Text to Text | 2 | ▲8.5%(up) | 297 | 27.4B |
| 6 | DeepSeek-V4.1-Flash-EXL3-2.8bpwjayleaton | Text Generation | 2 | — | 7 | 102B |
| 7 | LFM2.5-350M-BaseLiquidAI | Text Generation | 2 | ▲3.5%(up) | 36 | 354M |
| 8 | Qwen3.8-Flash-Next-exl3-CMP170HX-64GBmunekazu | Image Text to Text | 2 | — | 8 | 49.2B |
| 9 | Confucius4-R2T2-GGUFnetease-youdao | Speech Recognition | 2 | — | 28 | — |
| 10 | xlm-mlm-17-1280FacebookAI | Fill Mask | 2 | ▲0.1%(up) | 8 | — |
| 11 | Qwen3.6-35B-A3B-uncensored-heretic-Native-MTP-Preservedllmfan46 | Image Text to Text | 2 | ▲0.2%(up) | 47 | 35.1B |
| 12 | Swift1.5-Qwen3.8-Flash-Next-oQ4e-mtpscottlowry | — | 2 | — | 2 | 180B |
| 13 | Ministral-3-3B-Instruct-2512litert-community | Text Generation | 2 | ▲10.4%(up) | 5 | — |
| 14 | specter2_baseallenai | Embeddings | 2 | ▲2.3%(up) | 57 | — |
| 15 | mms-tts-kxmfacebook | Text to Speech | 2 | — | 3 | 36.3M |
| 16 | cvt-21-384microsoft | Image Classification | 2 | — | 3 | — |
| 17 | Devstral-2-123B-Instruct-2512mistralai | — | 2 | ▲0.5%(up) | 355 | 125B |
| 18 | bge-reranker-largeBAAI | Embeddings | 2 | ▲2.1%(up) | 484 | 560M |
| 19 | gemma-3-12b-it-norm-preserved-biprojected-abliteratedgrimjim | Image Text to Text | 2 | — | 31 | 12.2B |
| 20 | DiarizationLM-13b-Fisher-v1google | Text Generation | 2 | — | 18 | 13B |
| 21 | gemma-4-E4B-it-qat-mobile-ctgoogle | Any to Any | 2 | ▲2.0%(up) | 34 | 8.7B |
| 22 | gemma-4-E2B-it-qat-q4_0-unquantized-assistantgoogle | Any to Any | 2 | — | 27 | 78M |
| 23 | Llama-3.1-Nemotron-Nano-VL-8B-V1nvidia | Image Text to Text | 2 | ▲0.1%(up) | 188 | 8.7B |
| 24 | beit-large-patch16-512microsoft | Image Classification | 2 | ▲0.1%(up) | 20 | — |
| 25 | Phoenix-X-26B-A4B-Heretic-GGUFmradermacher | — | 2 | 0% | 2 | — |
| 26 | MIMESIS-4Bphanviethoang1512 | Text Generation | 2 | — | 3 | 5.2B |
| 27 | NigerianAccentedEnglishNCAIR1 | Speech Recognition | 2 | ▲0.8%(up) | 40 | — |
| 28 | CodeQwen1.5-7B-AWQQwen | Text Generation | 2 | ▲0.1%(up) | 9 | 7.3B |
| 29 | SmolVLM-256M-InstructHuggingFaceTB | Image Text to Text | 2 | ▲1.4%(up) | 423 | 256M |
| 30 | efficientnet-b3google | Image Classification | 2 | ▲1.1%(up) | 16 | — |
| 31 | vit-large-patch32-384google | Image Classification | 2 | ▲0.1%(up) | 22 | — |
| 32 | bekko-embedding-v1-a8mhotchpotch | Sentence Similarity | 2 | ▲17.6%(up) | 24 | 106M |
| 33 | Thomson-1.0-Small-APEX-I-MiniPlus-V2.1-GGUFIsValorum | Text Generation | 2 | — | 3 | — |
| 34 | Ternary-Bonsai-2-27B-Abliterated-v2-PQ2_0-MTP-GGUF-3060-benchgroxaxo | Text Generation | 2 | — | 3 | — |
| 35 | LFM2.5-24B-A2B-APEX-I-NanoPlus-GGUFIsValorum | Text Generation | 2 | — | 3 | — |
| 36 | Dayhoff-170M-GR-31000microsoft | Text Generation | 2 | — | 4 | 170M |
| 37 | tapex-large-finetuned-tabfactmicrosoft | Table Question Answering | 2 | ▲0.1%(up) | 15 | — |
| 38 | layoutlmv3-base-chinesemicrosoft | — | 2 | ▲0.6%(up) | 89 | — |
| 39 | Qwen3-235B-A22B-FP8Qwen | Text Generation | 2 | ▲0.3%(up) | 103 | 235B |
| 40 | MiniMax-H3-Looping-Sketch-AnimeInner-Reflections | — | 2 | — | 83 | — |
| 41 | conditional-detr-resnet-50microsoft | Object Detection | 2 | ▲1.8%(up) | 22 | 43.5M |
| 42 | llm-jp-4.1-8b-thinkingllm-jp | Text Generation | 2 | — | 16 | 8.6B |
| 43 | codebert-basemicrosoft | Embeddings | 2 | ▲0.1%(up) | 299 | — |
| 44 | laya-multilingual-GGUFmys | — | 2 | — | 15 | — |
| 45 | Qwen3-VL-8B-ComfyUInynxz | Image Text to Text | 2 | — | 4 | — |
| 46 | Mizan-27B-Turkish-Legal-LLM-Q4_K_M-GGUFAlicanKiraz0 | Image Text to Text | 2 | ▲1.2%(up) | 45 | — |
| 47 | gemma-4-12b-it-uncensored-GGUFzaakirio | Image Text to Text | 2 | ▲6.8%(up) | 72 | — |
| 48 | Qwen2.5-14B-Instruct-GPTQ-Int4Qwen | Text Generation | 2 | ▲3.0%(up) | 33 | 14.8B |
| 49 | VibeVoice-AcousticTokenizermicrosoft | Embeddings | 2 | ▲1.4%(up) | 20 | 687M |
| 50 | MiniMax-H3-FP8unsloth | Image Text To Video | 2 | — | 17 | — |