Scale AI's Argon claims first 60%+ image score, dominates long-video benchmarks
zacharynado · x · 2026-10-15
Scale AI Labs' new Argon model posted standout multimodal benchmark results: 62.8% on images, the first model to break 60% and roughly 7 points ahead of GPT-6-astra (55.6%), with Claude Opus 5.5 at 51.3%. The most striking result is long-video understanding — on clips over 10 minutes, Argon hits 59.1% while no other model gets past 41%.
More from Models
- vLLM ships Day-0 support for StepFun's Step-5 Preview: 604B MoE with 1M context — vllm_project · 2026-10-15
- vLLM's MTP3 validation: AIME25 up to 0.9667, prefill kernels up to 84% lower latency — vllm_project · 2026-10-15
- vLLM ships Day-0 support for Step 5 Preview: sparse attention, MoE, MTP and vision — vllm_project · 2026-10-15
- vLLM ships Day-0 support for Step 5 Preview with sparse attention, MoE and P/D on Blackwell — vllm_project · 2026-10-15
- Study: Grok cites 65.4 sources per answer, Gemini just 2.0 with 28% citation rate — drrickio · 2026-10-15
- OpenBMB ships MiniCPM-V 4.7 35B: beats Qwen3.6 on 24 of 34 benchmarks — petrusenko_max · 2026-10-15