ModelScope's banner

ModelScope

@ModelScope2022 • 15,020 subscribers

Driving innovations with open communities. 💬 Join our Discord: https://t.co/4M9AHVh5qa

Shorts

Introducing MiniMax-H3-Turbo from Wuli-Art. MiniMax-H3 audio-video generation now runs in as few as four steps. 🤖 ⚡ Choose from 4-step, 6-step, and 8-step variants for faster audio-video generation. 🎬 Supports text-to-video, first-frame and last-frame conditioning, with video and audio generated together. 🧩 The 4-step model uses a compact LoRA-based adapter, while the 6-step and 8-step versions provide complete Diffusers-format transformers. 📜 MiniMax-H3 Community License Agreement.

Introducing MiniMax-H3-Turbo from Wuli-Art. MiniMax-H3 audio-video generation now runs in as few as four steps. 🤖 ⚡ Choose from 4-step, 6-step, and 8-step variants for faster audio-video generation. 🎬 Supports text-to-video, first-frame and last-frame conditioning, with video and audio generated together. 🧩 The 4-step model uses a compact LoRA-based adapter, while the 6-step and 8-step versions provide complete Diffusers-format transformers. 📜 MiniMax-H3 Community License Agreement.

12,774 просмотров

Wan-Animate-2 pushes character animation beyond pose-driven pipelines: reference image + driving video in, cleaner character motion out. 🚀 🤖 🌐 🏆 Blind user study: preferred over Wan-Animate in 70%+ of overall-quality comparisons, with perceived quality comparable to Dreamina and Kling-MotionControl. 🎬 Cleaner animation: directly processes driving-video motion instead of relying on pose skeletons, reducing identity drift, artifacts, and shape distortion across different characters. 🎥 Text camera control: Viewpoint LoRA lets prompts steer the output perspective while keeping the original motion consistent. ⚡ Streaming path: Wan-Animate-2-Lite reports 24fps at 400x720, with stable long-sequence generation for live avatars and interactive virtual environments. Built with dual-branch DiT, Time-Align RoPE, Sparse-Ref Attention, 100K+ video pairs, and 50K Unreal Engine multi-view samples.

Wan-Animate-2 pushes character animation beyond pose-driven pipelines: reference image + driving video in, cleaner character motion out. 🚀 🤖 🌐 🏆 Blind user study: preferred over Wan-Animate in 70%+ of overall-quality comparisons, with perceived quality comparable to Dreamina and Kling-MotionControl. 🎬 Cleaner animation: directly processes driving-video motion instead of relying on pose skeletons, reducing identity drift, artifacts, and shape distortion across different characters. 🎥 Text camera control: Viewpoint LoRA lets prompts steer the output perspective while keeping the original motion consistent. ⚡ Streaming path: Wan-Animate-2-Lite reports 24fps at 400x720, with stable long-sequence generation for live avatars and interactive virtual environments. Built with dual-branch DiT, Time-Align RoPE, Sparse-Ref Attention, 100K+ video pairs, and 50K Unreal Engine multi-view samples.

26,091 просмотров

Want better MiniMax-H3 LoRAs? DiffSynth-Studio just open-sourced two things worth checking out. 🔥 ⚙️ The MiniMax-H3 Training Adapter, a rank 64 LoRA. Combined with differential training, it helps you train higher quality LoRAs: 🔗 📦 The self-generated dataset behind the adapter, fully open as well: 🔗 One more thing: the adapter will soon power MiniMax-H3 LoRA training on ModelScope Civision. Stay tuned. 🚀

Want better MiniMax-H3 LoRAs? DiffSynth-Studio just open-sourced two things worth checking out. 🔥 ⚙️ The MiniMax-H3 Training Adapter, a rank 64 LoRA. Combined with differential training, it helps you train higher quality LoRAs: 🔗 📦 The self-generated dataset behind the adapter, fully open as well: 🔗 One more thing: the adapter will soon power MiniMax-H3 LoRA training on ModelScope Civision. Stay tuned. 🚀

10,437 просмотров

Videos

Anya Rossi

sweetdream.ai

SweetDream.ai•Sponsored•Livecam

Watch Anya Live

Anya is streaming live right now! Join her private show and enjoy exclusive content.

Exclusive private shows

1.2k viewers online

Private Show

Join now for exclusive access

Free preview available • Premium content

Meet LingBot-World 2.0.🌍A 1.3B world model, running continuous interactive generation in real time on a single consumer GPU. 🤖 ⚡ It responds to movement, camera control, combat, and text-driven events in real time, while the scene keeps unfolding instead of ending as a fixed video clip. 🌌 Causal pretraining and MoBA help preserve textures, geometry, and scene consistency over long autoregressive rollouts. 🧠 Consistency and distribution-matching distillation compress multi-step generation and train the model on its own rollouts, reducing accumulated drift. 🛠️ The 14B Causal Pretrain backbone and Bidirectional teacher are also open, giving developers a path to post-train, distill, compress, and adapt new world models. 🔒 CC BY-NC-SA 4.0. Non-commercial use only.

Meet LingBot-World 2.0.🌍A 1.3B world model, running continuous interactive generation in real time on a single consumer GPU. 🤖 ⚡ It responds to movement, camera control, combat, and text-driven events in real time, while the scene keeps unfolding instead of ending as a fixed video clip. 🌌 Causal pretraining and MoBA help preserve textures, geometry, and scene consistency over long autoregressive rollouts. 🧠 Consistency and distribution-matching distillation compress multi-step generation and train the model on its own rollouts, reducing accumulated drift. 🛠️ The 14B Causal Pretrain backbone and Bidirectional teacher are also open, giving developers a path to post-train, distill, compress, and adapt new world models. 🔒 CC BY-NC-SA 4.0. Non-commercial use only.

33,630 просмотров • 1 месяц назад

NeoHorse-Jev-4B is now open—a compact model built to turn application states directly into structured decisions and probabilities. 🤖 🏆 Scores 77.70 across six text decision benchmark groups, ranking first among the four open-weight models with complete results in the comparison. ⚡ Prefill-only inference avoids autoregressive answer generation and supports three decision primitives: Choice for action selection, Noul for yes/no judgment, and Score for ordered ratings. 🎮 The same interface can power request routing, tool selection, workflow control, games, robot manipulation, and autonomous-driving simulations. 🖼️ Supports text inputs or a single image combined with text, returning normalized probabilities that applications can use directly. 🛠️ Deploy locally through vLLM, SGLang, Python, CLI, or HTTP. Apache 2.0.

NeoHorse-Jev-4B is now open—a compact model built to turn application states directly into structured decisions and probabilities. 🤖 🏆 Scores 77.70 across six text decision benchmark groups, ranking first among the four open-weight models with complete results in the comparison. ⚡ Prefill-only inference avoids autoregressive answer generation and supports three decision primitives: Choice for action selection, Noul for yes/no judgment, and Score for ordered ratings. 🎮 The same interface can power request routing, tool selection, workflow control, games, robot manipulation, and autonomous-driving simulations. 🖼️ Supports text inputs or a single image combined with text, returning normalized probabilities that applications can use directly. 🛠️ Deploy locally through vLLM, SGLang, Python, CLI, or HTTP. Apache 2.0.

19,925 просмотров • 29 дней назад

Say hello to DreamX-Creator 1.0, a 7B native audio-video generator with local 2K refinement. Apache 2.0. 🤖 🏆 Delivers results comparable to much larger SOTA models like MiniMax H3 on Verse Bench, reaching 0.1351 DeSync after RL and 0.6930 VQ with the 2K Refiner. 🎬 Turn a first frame and prompt into synchronized video, dialogue, action sounds, ambience, weather, crowds, and music. No separate dubbing stage. 🔊 Gated Cross-Modal Attention lets sound and visuals shape each other, while modality-aware RL improves video, audio, and their alignment. ✨ The one-step 2K Refiner tops the compared refinement methods on MUSIQ (0.7073) and MANIQA (0.4382) while preserving motion and audio timing.

Say hello to DreamX-Creator 1.0, a 7B native audio-video generator with local 2K refinement. Apache 2.0. 🤖 🏆 Delivers results comparable to much larger SOTA models like MiniMax H3 on Verse Bench, reaching 0.1351 DeSync after RL and 0.6930 VQ with the 2K Refiner. 🎬 Turn a first frame and prompt into synchronized video, dialogue, action sounds, ambience, weather, crowds, and music. No separate dubbing stage. 🔊 Gated Cross-Modal Attention lets sound and visuals shape each other, while modality-aware RL improves video, audio, and their alignment. ✨ The one-step 2K Refiner tops the compared refinement methods on MUSIQ (0.7073) and MANIQA (0.4382) while preserving motion and audio timing.

31,067 просмотров • 1 месяц назад

Only ~1.2B parameters active at a time. Edge0-8B-A1B-preview makes an 8B-class MoE practical for local inference.📜 Apache 2.0. 🤖 ⚡ Reaches 23.9–25.3 tokens/s with about 1.0 GiB peak active memory in the reported short-context benchmark. 🏆 Retains most of the FP16 base model’s quality, with an average gap of just 2.8 points across five benchmarks. MMLU-Pro rises from 65.8 to 70.1. 🧠 SSD expert offload keeps most weights outside active memory, while a prerouter predicts which experts each layer will need next. 🛠 Recover-LoRA offsets quality loss from 4-bit quantization and expert offloading. The current Preview runs locally through MLX on Apple Silicon.

Only ~1.2B parameters active at a time. Edge0-8B-A1B-preview makes an 8B-class MoE practical for local inference.📜 Apache 2.0. 🤖 ⚡ Reaches 23.9–25.3 tokens/s with about 1.0 GiB peak active memory in the reported short-context benchmark. 🏆 Retains most of the FP16 base model’s quality, with an average gap of just 2.8 points across five benchmarks. MMLU-Pro rises from 65.8 to 70.1. 🧠 SSD expert offload keeps most weights outside active memory, while a prerouter predicts which experts each layer will need next. 🛠 Recover-LoRA offsets quality loss from 4-bit quantization and expert offloading. The current Preview runs locally through MLX on Apple Silicon.

16,923 просмотров • 1 месяц назад

An agent model, plus two multimodal models that can see and operate computers. Nex-AGI releases Nex-N2.5 in mini, Pro, and Max sizes.📜 Apache 2.0. 🤖 🏆 Max leads the published BrowseComp comparison with 92.6. Pro ranks first on OSWorld-G with 87.4. 🖥️ The 35B mini and 397B Pro browse the web, operate software, verify results, and self-correct through visual feedback. 🧠 The 1.6T Max targets complex text-based agent tasks, from coding and tool use to research and knowledge work. It marks Nex-AGI’s first full post-training run at trillion-parameter scale. ⚙️ All three support function calling and direct, adaptive, or full thinking modes.

An agent model, plus two multimodal models that can see and operate computers. Nex-AGI releases Nex-N2.5 in mini, Pro, and Max sizes.📜 Apache 2.0. 🤖 🏆 Max leads the published BrowseComp comparison with 92.6. Pro ranks first on OSWorld-G with 87.4. 🖥️ The 35B mini and 397B Pro browse the web, operate software, verify results, and self-correct through visual feedback. 🧠 The 1.6T Max targets complex text-based agent tasks, from coding and tool use to research and knowledge work. It marks Nex-AGI’s first full post-training run at trillion-parameter scale. ⚙️ All three support function calling and direct, adaptive, or full thinking modes.

18,509 просмотров • 1 месяц назад

Breeze-TTS-2 is here. 🎙️ Real-time voice cloning, design, and direction in one model. 🤖 🏆 #1 open-weight TTS on Artificial Analysis, plus first place on voice design, instruction following, and latency benchmarks. 🎭 Clone a voice while preserving timbre, rhythm, emotion, and style, or design and direct new Chinese and English voices with natural-language prompts and inline vocal events. ⚡ Under 40 ms TTFA and 3.1× real-time streaming on the warmed-up fast path. Eager inference uses ~7.7 GiB, with a 12 GB GPU recommended. 📜 Code: Apache 2.0. Weights, derivatives, and self-hosted outputs: research and non-commercial use only.

Breeze-TTS-2 is here. 🎙️ Real-time voice cloning, design, and direction in one model. 🤖 🏆 #1 open-weight TTS on Artificial Analysis, plus first place on voice design, instruction following, and latency benchmarks. 🎭 Clone a voice while preserving timbre, rhythm, emotion, and style, or design and direct new Chinese and English voices with natural-language prompts and inline vocal events. ⚡ Under 40 ms TTFA and 3.1× real-time streaming on the warmed-up fast path. Eager inference uses ~7.7 GiB, with a 12 GB GPU recommended. 📜 Code: Apache 2.0. Weights, derivatives, and self-hosted outputs: research and non-commercial use only.

16,414 просмотров • 1 месяц назад

Listen, speak, handle interruptions, and call tools in one live conversation. 🎙️ Introducing NVIDIA-NemotronLabs-VoiceChat-11B, NVIDIA’s end-to-end full-duplex model for real-time voice agents. 🤖 🏆 It ranks #2 among open full-duplex models on both VoiceBench and Full-Duplex-Bench 1.0. ⚡ Natural turn-taking responds in ~448 ms, while barge-in lets users interrupt the model with ~480 ms latency. 🛠️ The first open full-duplex model with live tool calling. One unified system handles speech understanding, generation, transcription, and tool execution while keeping the conversation flowing. 🧠 Built with a Fast Conformer, Nemotron Nano v2, and a TTS decoder, trained on ~550K hours of real and synthetic speech. 📜 Research use only. OpenMDW 1.1.

Listen, speak, handle interruptions, and call tools in one live conversation. 🎙️ Introducing NVIDIA-NemotronLabs-VoiceChat-11B, NVIDIA’s end-to-end full-duplex model for real-time voice agents. 🤖 🏆 It ranks #2 among open full-duplex models on both VoiceBench and Full-Duplex-Bench 1.0. ⚡ Natural turn-taking responds in ~448 ms, while barge-in lets users interrupt the model with ~480 ms latency. 🛠️ The first open full-duplex model with live tool calling. One unified system handles speech understanding, generation, transcription, and tool execution while keeping the conversation flowing. 🧠 Built with a Fast Conformer, Nemotron Nano v2, and a TTS decoder, trained on ~550K hours of real and synthetic speech. 📜 Research use only. OpenMDW 1.1.

22,605 просмотров • 2 месяцев назад

MiniMax-Music3 is now open source. 🎶 Generate complete songs up to five minutes long, with expressive vocals, evolving arrangements, and coherent structure from intro to outro. 🤖 🧠 Around 11.1B parameters: an 8B Global LLM for long-range structure, a 0.6B Local LLM for acoustic detail, plus a 2.4B Flow Matching model and 123M decoder. 🎤 Control lyrics, song sections, genre, mood, vocals, instrumentation, and arrangement development. 🎧 Generates 32 kHz, 16-bit stereo WAV audio with stable long-form quality. Listen to the demos👇

MiniMax-Music3 is now open source. 🎶 Generate complete songs up to five minutes long, with expressive vocals, evolving arrangements, and coherent structure from intro to outro. 🤖 🧠 Around 11.1B parameters: an 8B Global LLM for long-range structure, a 0.6B Local LLM for acoustic detail, plus a 2.4B Flow Matching model and 123M decoder. 🎤 Control lyrics, song sections, genre, mood, vocals, instrumentation, and arrangement development. 🎧 Generates 32 kHz, 16-bit stereo WAV audio with stable long-form quality. Listen to the demos👇

18,883 просмотров • 2 месяцев назад

YuE2-3B turns lyrics into complete songs through an editable musical score. 🤖 📄 🏆 YuE2 best-of-8 reaches 6.9632 SongBench Avg on WildSongBench, the highest score across the evaluated open and proprietary systems. Suno v5 scores 6.8721. 🎼 It plans melody and chords before generating vocals and accompaniment. Edit the score or supply your own ABC notation. 🎙️ The same checkpoint supports direct song generation, zero-shot covers, and agent-guided edits to the score, style, and lyrics. ⚡ Generate 48 kHz stereo audio locally on a 24GB GPU without quantization. 📜 Weights: CC BY-NC 4.0. Code: Apache 2.0.

YuE2-3B turns lyrics into complete songs through an editable musical score. 🤖 📄 🏆 YuE2 best-of-8 reaches 6.9632 SongBench Avg on WildSongBench, the highest score across the evaluated open and proprietary systems. Suno v5 scores 6.8721. 🎼 It plans melody and chords before generating vocals and accompaniment. Edit the score or supply your own ABC notation. 🎙️ The same checkpoint supports direct song generation, zero-shot covers, and agent-guided edits to the score, style, and lyrics. ⚡ Generate 48 kHz stereo audio locally on a 24GB GPU without quantization. 📜 Weights: CC BY-NC 4.0. Code: Apache 2.0.

11,401 просмотров • 1 месяц назад

A real-time world simulator, now in 0.5B. ABot-World-0.5B-LF brings action-conditioned interactive rollout to ModelScope. 🎮License: Apache 2.0. 🤖 ⚡ Desktop-ready rollout: reported at 720p, 16 FPS, 1.2s latency, and 19GB GPU memory on a single-desktop setup. 🌍 Infinite interaction: responds to user actions continuously and keeps generating beyond fixed video-length limits. 🎬 LongForcing training helps the model expand new scenes and dynamics during rollout, without prompt switching or scene lock-in. 🛠️ Built on Wan2.2-TI2V-5B as a causal student model for lower-latency interactive world rollout.

A real-time world simulator, now in 0.5B. ABot-World-0.5B-LF brings action-conditioned interactive rollout to ModelScope. 🎮License: Apache 2.0. 🤖 ⚡ Desktop-ready rollout: reported at 720p, 16 FPS, 1.2s latency, and 19GB GPU memory on a single-desktop setup. 🌍 Infinite interaction: responds to user actions continuously and keeps generating beyond fixed video-length limits. 🎬 LongForcing training helps the model expand new scenes and dynamics during rollout, without prompt switching or scene lock-in. 🛠️ Built on Wan2.2-TI2V-5B as a causal student model for lower-latency interactive world rollout.

19,642 просмотров • 3 месяцев назад

🤖 Introducing InternVLA-A1 — now fully open-sourced! Many VLA models follow instructions well in static scenes… but struggle in dynamic environments (conveyor belts, rotating platforms, multi-robot setups). Why? They see the present—but can’t imagine the future. InternVLA-A1 solution: unify perception, imagination, and action in one model: ✅ Scene understanding: Image + text → task parsing ✅ Task imagination: Predict future frames → reason about dynamics ✅ Guided control: Execute actions steered by visual foresight Powered by InternData-A1 - Large-scale high-quality simulated dataset, InternVLA-A1 stays robust under complex backgrounds, lighting, and distractions. 🔥 See it in action: 1️⃣ High-speed conveyor: track, predict, and stably grasp or flip packages 2️⃣ Rotating platform: task-aware recognition & precise pick-up of diverse items 📊 Outperforms π0 and Gr00t N1.5 on general manipulation benchmarks! ✨ Model, data, and code are all open! Models: Datasets: GitHub:

🤖 Introducing InternVLA-A1 — now fully open-sourced! Many VLA models follow instructions well in static scenes… but struggle in dynamic environments (conveyor belts, rotating platforms, multi-robot setups). Why? They see the present—but can’t imagine the future. InternVLA-A1 solution: unify perception, imagination, and action in one model: ✅ Scene understanding: Image + text → task parsing ✅ Task imagination: Predict future frames → reason about dynamics ✅ Guided control: Execute actions steered by visual foresight Powered by InternData-A1 - Large-scale high-quality simulated dataset, InternVLA-A1 stays robust under complex backgrounds, lighting, and distractions. 🔥 See it in action: 1️⃣ High-speed conveyor: track, predict, and stably grasp or flip packages 2️⃣ Rotating platform: task-aware recognition & precise pick-up of diverse items 📊 Outperforms π0 and Gr00t N1.5 on general manipulation benchmarks! ✨ Model, data, and code are all open! Models: Datasets: GitHub:

38,118 просмотров • 9 месяцев назад

Introducing LingBot-World: An open-source world simulator pushing the boundaries of video generation. 🚀 🌍 High-Fidelity: Realistic, scientific, & stylized. 🧠 Long-Term Memory: Minute-level consistency. ⚡ Real-Time: <1s latency at 16 FPS. 📜 Apache 2.0 Licensed. Model: Github:

Introducing LingBot-World: An open-source world simulator pushing the boundaries of video generation. 🚀 🌍 High-Fidelity: Realistic, scientific, & stylized. 🧠 Long-Term Memory: Minute-level consistency. ⚡ Real-Time: <1s latency at 16 FPS. 📜 Apache 2.0 Licensed. Model: Github:

28,809 просмотров • 8 месяцев назад

MOSS-TTS v1.5 is here, an upgrade to v1.0 from @OpenMOSS. (demo👇)🤖 Key improvements: ⏸️ Inline pause control: [pause 3.2s] now supported mid-sentence 🌍 31 languages, up from 20 — now includes Cantonese, Hindi, Thai, Vietnamese, Tagalog, Swahili and more 🎙️ More stable voice cloning with reduced variance across repeated generations 📝 Better long-reference, short-text cloning All v1.0 capabilities preserved: zero-shot cloning, long-form speech, Pinyin/IPA control, code-switching. 💻

MOSS-TTS v1.5 is here, an upgrade to v1.0 from @OpenMOSS. (demo👇)🤖 Key improvements: ⏸️ Inline pause control: [pause 3.2s] now supported mid-sentence 🌍 31 languages, up from 20 — now includes Cantonese, Hindi, Thai, Vietnamese, Tagalog, Swahili and more 🎙️ More stable voice cloning with reduced variance across repeated generations 📝 Better long-reference, short-text cloning All v1.0 capabilities preserved: zero-shot cloning, long-form speech, Pinyin/IPA control, code-switching. 💻

13,921 просмотров • 5 месяцев назад

🚀 New on ModelScope: MiniMax M2.1 is open-source! ✅ SOTA in 8+ languages (Rust, Go, Java, C++, TS, Kotlin, Obj-C, JS) ✅ Full-stack Web & mobile dev: Android/iOS, 3D visuals, vibe coding that actually ships ✅ Smarter, faster, 30% fewer tokens — with lightning mode (M2.1-lightning) for high-TPS workflows ✅ Top-tier on SWE-bench, VIBE, and custom coding/review benchmarks ✅ Works flawlessly in Cursor, Cline, Droid, BlackBox, and more It’s not just “better code” — it’s AI-native development, end to end. 🔗 Model:

🚀 New on ModelScope: MiniMax M2.1 is open-source! ✅ SOTA in 8+ languages (Rust, Go, Java, C++, TS, Kotlin, Obj-C, JS) ✅ Full-stack Web & mobile dev: Android/iOS, 3D visuals, vibe coding that actually ships ✅ Smarter, faster, 30% fewer tokens — with lightning mode (M2.1-lightning) for high-TPS workflows ✅ Top-tier on SWE-bench, VIBE, and custom coding/review benchmarks ✅ Works flawlessly in Cursor, Cline, Droid, BlackBox, and more It’s not just “better code” — it’s AI-native development, end to end. 🔗 Model:

16,939 просмотров • 10 месяцев назад

Больше нет контента для загрузки