Open-weights model ranks in top 3 of AI video Elo ranking as native audio becomes standard
MarkTechPost's AI video generation ranking, dated Nov 13, 2026, uses blind human-preference Elo from the Artificial Analysis Video Arena. An open-weights model now sits in the top 3, and native audio is described as table stakes.
An open-weights model now sits in the top 3 of MarkTechPost's ranking of AI video generation models, published Nov 13, 2026. The outlet says the leaderboard looks very different from 12 months ago.
What the ranking uses
The ranking, as of Nov 13, 2026, uses one quality signal: blind human-preference Elo from the Artificial Analysis Video Arena. MarkTechPost's article title lists three comparison axes: Elo, price per second and open weights.
Changes MarkTechPost highlights
According to MarkTechPost, three shifts separate the current leaderboard from a year ago:
- Native audio is now table stakes. Audio generated alongside video is described as a baseline expectation rather than a differentiator.
- Clips run 15 to 30 seconds. MarkTechPost reports this as the current range of clip length.
- An open-weights model sits in the top 3. The ranking places a model with openly available weights among the three highest-rated systems.
What this means
The ranking relies on blind human-preference Elo, so its ordering reflects how viewers judge outputs side by side rather than a fixed technical benchmark. By MarkTechPost's account, an open-weights model in the top 3 means teams that want to run video generation themselves are no longer limited to lower-ranked options on this leaderboard. With native audio described as standard, a video model without it is now measured against a baseline that includes sound.
Related Articles
Gemini 4 Hydrogen scores 54 on Artificial Analysis Intelligence Index, just below Claude Sonnet 5.5 and GPT-6.1 Sol
Artificial Analysis reports that Google's Gemini 4 Hydrogen scores 54 on its Intelligence Index, just below Claude Sonnet 5.5 and GPT-6.1 Sol. It costs a third of Sonnet's cost but about 2.5x that of GPT-6.1 Sol, and adds an xhigh reasoning level.
NVIDIA serves 753B-parameter GLM-5.3 NVFP4 on two DGX Stations, peaking at 607.8 tokens/s
NVIDIA published a tutorial for serving GLM-5.3 NVFP4, a 753-billion-parameter MoE model, across two GB300 DGX Stations with vLLM. In NVIDIA's AIPerf tests, output throughput peaked at 607.8 tokens per second at concurrency 16 and fell to 385.0 at concurrency 32.
Ai2's XYBench: Five LLMs Stay Focused on Users' Mistaken Approach Instead of Their Real Goal
Ai2 released XYBench, a benchmark of 8,115 real-world questions that tests whether AI assistants recognize when a user's request rests on a misconception. Five evaluated models, including Claude Sonnet 4.6 and GPT-5, were less likely than human reference answers to steer users toward their actual goal. The paper won the Best Resource Paper Award at EMNLP 2026.
Future Doctor's MedGPT beats 250 hospital physicians 88 to 12 on 40 complex lung cancer cases
Chinese medical AI company Future Doctor (未来医生) and thoracic surgery teams at top-tier hospitals report that its MedGPT won 88 to 12 against 250 physicians on 40 unpublished complex lung cancer cases. MedGPT scored 96.22 in a head-to-head comparison with general-purpose models, against 71.38 for GPT-5.4, 61.81 for Claude Opus 4.7 and 54.62 for Gemini 3.1 Pro.
Comments
Loading...