KVARK Sparkle 1.0 open-weights: Qwen3.8-27B derivative cuts thinking tokens by up to 67%
dzulajla98 · reddit · 2026-11-02
Egzakta AI Lab released KVARK Sparkle 1.0 (27B), an Apache 2.0 open-weight model based on Qwen3.8-27B focused on reasoning efficiency.
- Up to 67% fewer thinking tokens; median response time 2× faster (13.7s vs 28.1s)
- Comparable accuracy on GPQA Diamond, AIME, LiveCodeBench, IFEval and MMLU-Pro, with no statistically significant differences
- Practical tasks: dashboard generation 1.6× faster, landing pages 1.4×, pelican-on-a-bicycle SVG 3.2×
- Ships in BF16, FP8 and GGUF; supports llama.cpp, vLLM, Transformers and SGLang
More from Models
- Mistral employee shows its model autonomously picked STO-3G basis set, hit 1e-13 accuracy vs PySCF — MistralAI · 2026-11-02
- Fable 5.5 launch week teased: "see you real soon" — iruletheworldmo · 2026-11-02
- Amazon CEO expects just 6-8 AI models to be 'good enough' for most enterprise work — YvesMulkers · 2026-11-02
- ZenteiQ, backed by IndiaAI Mission, to unveil homegrown foundation models tomorrow — deepakns · 2026-11-02
- Altman: model that cracked Navier-Stokes is the next GPT, run as 10,000 agents for 4 days — Hesamation · 2026-11-02
- Tesla FSD Nears 17 Billion Miles, Adding 1.8M Miles Per Hour — XFreeze · 2026-11-02