Same-prompt test: GPT-6 Astra outdetails Gemini 4 in scene generation
VraserX · x · 2026-09-19
A same-prompt comparison of Gemini 4 vs GPT-6 Astra's scene/gameplay generation found GPT-6 Astra produces far more detailed characters and environments, while Gemini 4 delivers a more basic gameplay scene. The comparer crowns GPT-6 Astra but notes Gemini 4 is still decent; reposter VraserX quips Google may be in trouble if this is all Gemini 4 offers. Single-prompt test, not a systematic eval.
More from Models
- Codex usage limits finally reset after days of users hitting exhausted quotas — CtrlAltDwayne · 2026-09-19
- Rumored Opus 5.2 generates a full 5-minute Titanic film in pure JavaScript with three.js — dotey · 2026-09-19
- Model Remembers 'cranberry-42' Across App Restarts in Persistent-Session Test — RileyRalmuto · 2026-09-19
- Free models aren't the real problem: studies show hallucinations persist in SOTA LLMs — AryHHAry · 2026-09-19
- Why do AI models lack personality? Safety training and attachment fears, explained — alexisgallagher · 2026-09-19
- New model JEV is free on OpenRouter, Vercel, Cloudflare and others, at ~$0.042/M input tokens — airesearch12 · 2026-09-19