New Model Spark-X2.5: 4B variant matches Qwen 3.5 9B
insraq · reddit · 2026-09-01
A new model series Spark-X2.5 (4B/1.7B params) appeared on Hugging Face, featuring a custom architecture rather than a standard fine-tune.
- Performance: The 4B variant claims benchmark scores neck and neck with Qwen 3.5 9B.
- Support: It doesn't run out-of-the-box on llama.cpp yet; a pending PR and a custom fork are available.
The community is discussing early experiences with the model.
More from Models
- MOSS SFT vs Flux 3 Klein vs SD 3.0: Same-Prompt Comparison Exposes SD3's Age — ZootAllures9111 · 2026-10-18
- Quick take: GPT-6 matches Opus 5.5 at coding, far ahead at computer use — _vztu · 2026-10-18
- Open models now above Sol-level, outpacing expectations as community awaits V4.1 Pro — yacineMTB · 2026-10-18
- Codex took 3 days, Opus 5.5 nailed it in an afternoon — banned user's video protest — dotey · 2026-10-18
- Yacine MTB: Users will remember how Anthropic treated them once intelligence proliferates — yacineMTB · 2026-10-18
- Mistral's Timestral forecasting models start at 5M params and run on plain CPUs — lmoroney · 2026-10-18