Llama 3.3 Nemotron 49B Super appears on LMSYS Arena
Illustrious asking people to pay $371,000 (discounted price) for releasing Illustrious v3.5 Vpred.
It seems OnomaAI raised the funding goal of Illustrious 3.0 to 150k dollars and the goal of 3.5 v-pred to 530k dollars.
[Kohya news] wan 25% speed up | Release of Kohya's work following the legendary Kohya Deep Shrink
Sharing my build: Budget 64 GB VRAM GPU Server under $700 USD
5090 Secured! Need CPU Advice for Local LLMs vs. 9950X3D/9800X3D
Illustrious v3.5-pred is already trained and has raised 100% Stardust, but they will not open the model weights (at least not for 300,000 Stardust).
Skip layer guidance has landed for wan video via KJNodes
DeepSeek's owner asked R&D staff to hand in passports so they can't travel abroad. How does this make any sense considering Deepseek open sources everything?
Qwen release next week will be "smaller". Full release of QwQ-Max "a little bit later"
Nunchaku v0.1.4 released!
LTXV vs. Wan2.1 vs. Hunyuan – Insane Speed Differences in I2V Benchmarks!
Get a 30% speed improvement on Wan with the AdaptiveGuider node.
Wan2.1 reminds me of the first release of SD 1.5, It's underrated, one of the biggest gifts we received IMO since SD1.5.
Wan VS Hunyuan
Tencent Releases HunyuanVideo-I2V: A Powerful Open-Source Image-to-Video Generation Model
[UPDATE] Instead of training 100 Hunyuan Video LoRAs, I am launching a Wan 2.1 T2V Generator and started training LoRAs on Wan 14B
LTX-Video v0.9.5 released, now with keyframes, video extension, and higher resolutions support.
Qwen/QwQ-32B · Hugging Face
Hunyuan I2V release date
WAN Released
WanX - Alibaba is about open-source this model - Hope it fits consumer GPUs
Drummer's Skyfall 36B v2 - An upscale of Mistral's 24B 2501 with continued training; resulting in a stronger, 70B-like model!
Zonos: Incredible new TTS model from Zyphra
hunyuan-image2video V2 update
