analiza aiqbench
Kling 4.0 is a product launch from a generative video model, and the most useful details for practitioners are the concrete specs: up to 30 seconds of native output per pass, 10-bit HDR, 4K and 21:9 support, and claimed gains in stability for high-motion scenes and complex camera moves. Those are self-reported capabilities, so independent evaluations of temporal consistency, motion artifacts, and prompt adherence will matter more than the headline numbers. Readers tracking video generation should watch whether the single-pass duration holds up in real workflows.
Kling AI Launches Kling 4.0, Supporting 10-bit HDR and 30-Second Native Output, Setting a New Benchmark for Video Models
According to 36Kr, Kling AI launched Kling 4.0, a new generation of its video generation model, on October 15. In terms of core capabilities, the model can natively generate up to 30 seconds in a single pass (with text input of up to 8,000 tokens), supports up to 10-bit HDR and 4K and 21:9 ultra-wide aspect ratio output, and has improved the stability of its performance with high-dynamic and complex camera movements.
Citește articolul complet pe 36KrSubiecte conexe
SPIR-V on ROCm: A Portable IR for AMD GPUs
AMD · 20 iul. 2026

[AINews] not much happened today
Latent Space · 3 oct. 2026
Benchmarking Kimi-K3 Across vLLM, SGLang, and ATOM on MI350X
AMD · 22 sept. 2026

[AINews] Opus 5.5 is good at explainer videos
Latent Space · 29 sept. 2026