ASCIITermDraw-Bench tests whether VLMs can draw ASCII diagrams
East-Muffin-6472 · reddit · 2026-07-20
ASCIITermDraw-Bench is a new benchmark for testing whether vision-language models can generate and edit ASCII diagrams. The authors argue that ASCII can be useful for communicating architectures, topologies, and cluster layouts, and that this capability is different from simply describing a diagram. The benchmark contains 80 tasks across four areas: basic box/layout drawing, network topologies, software architecture diagrams, and image-conditioned diagram editing. Each answer is scored with both a structural metric and an LLM-based semantic judge repeated five times, with a 95% confidence interval reported. The current leaderboard is headed by Gemma-4-31B-IT, followed by Qwen3.7-Plus, Kimi-K2.6, MiniMax-M3, Qwen3.5-9B, and Ternary-Bonsai-27B.
Related event: ASCIITermDraw-Bench Tests VLMs on ASCII Drawing(2 posts)→
More from Multimodal
- Short film 'Odissey of Aeons' made with MiniMax H3 video generation — Musikalandia · 2026-10-27
- Remiqora Video: open-source local studio turns one idea into 4K chained AI videos — inikolax · 2026-10-27
- Single HTML/JS file generates creative art and copy in-browser with zero LLM calls — Afinetheorem · 2026-10-27
- $5k kale-img-1 fine-tuning contest won with MOSS-Image base, SDXL trails far behind — ostrisai · 2026-10-27
- HN discussion: How far image generation has come since August 2022 — hunterbrooks · 2026-10-27
- FLUX.3 Edit nails a -90° car rotation on first try with just two reference images — EvilBot-666 · 2026-10-27