DeepSeek V4 Vision vs OX-Alpha: Multimodal Agent Gaming Benchmark
karminski3 · x · 2026-08-23
The author benchmarks the newly released DeepSeek-V4-Flash-Vision-Exp against the anonymous model OX-Alpha on OpenRouter using a custom AI esports coaching framework. To ensure fairness, both models process video frames since DeepSeek's multimodal capabilities do not support direct video input. The test evaluates their ability to act as agents by accurately analyzing CS2 gameplay footage, identifying player strengths and weaknesses, and suggesting improvements. The video also details the optimal image input ratios for both models.
More from coding & agent
- Engineer: my job went from silently thinking at a computer to constantly talking to it — KevinNaughtonJr · 2026-10-16
- Vercel Goes Full Backend: Python GA, Containers, and Daemons — cramforce · 2026-10-16
- WorldMemArena Accepted at NeurIPS: Memory Storage ≠ Memory Use — xwang_lk · 2026-10-16
- Vercel Ship Demos Container Deploys for Go, Ruby, Java and More — edwin · 2026-10-16
- TRL adds day-0 support for fine-tuning Muse Spark Open's 816B MoE with LoRA on 64 H100s — QGallouedec · 2026-10-16
- Databricks engineers land impactful PRs on day 2 with AI dev tools, vs week 3 before — Yuchenj_UW · 2026-10-16