New benchmark just dropped!
Summary
A new benchmark compares the performance of Qwen3.8-27b, Sol 5.6, and Qwen3.6-35B models on generating an SVG of a horse on a bicycle in the desert with a camel background, using a typo-included prompt.
Similar Articles
Qwen3.8 27b just exceeded my expectations on svg generation :D
A user tested the Qwen3.8 27B model on SVG generation with a complex prompt and found the results exceeded expectations, highlighting the model's capability in creative tasks.
Qwen3.6-35B-A3B on my laptop drew me a better pelican than Claude Opus 4.7
Simon Willison compares Qwen3.6-35B-A3B running locally on a MacBook Pro against Claude Opus 4.7, finding that Qwen produces better SVG illustrations of pelicans riding bicycles and flamingos on unicycles, though he notes this narrow benchmark doesn't reflect broader model capabilities.
I just ran Qwen 3.8 27 in Q4 against GPT 5.6 Sol high - and it easily won against SOL - complex animated SVG tasks
A user tested Qwen 3.8 27B in Q4 quantization against GPT 5.6 SOL in complex animated SVG tasks and found that Qwen performed better with fewer errors, highlighting strengths in spatial reasoning and coding.
Forget the Pelican, it's Weevil-Time! / Benchmaxxing-Proof SVG and Vision Benchmark
The article describes testing the Qwen3.8-27B AI model with different quantizations and settings to recreate images as SVG, aiming to develop a benchmark resistant to benchmaxxing. Preliminary results indicate that high reasoning effort and specific cache configurations optimize performance.
Horsey challenge | Strata Fork | Qwen 3.8 FN NVFP4
用户在 RTX 5090 单卡上使用 strata-nvfp4 引擎（Strata 的 NVFP4 分支）运行 Qwen3.8-Flash-Next 125B-A6B 混合 MoE 模型，分享了「小马挑战」（生成 Three.js 低多边形奔跑小马的 HTML）的性能数据：87 步、12.4 万输出 token、平均 103 t/s 解码速度、99% 缓存命中率。