Nemotron 3.5 Bolt, Muse Spark Open, and a Fast LFM2.5 24B
Nemotron 3.5, with a closer look at Bolt and its “native 2-bit” quantization
Proof1 independent outlet1 community thread
Key points
- Muse Spark Open: finally released, but not quite what I was hoping for
- Nemotron 3.5: Bolt Deserves Some Special Attention
- Nemotron 3.5 Bolt is a 74B-A9B model compressed from Super and quantized for a roughly 24 GB footprint.
- That makes it a useful alternative to Qwen3.6 35B, and maybe it will stay like this for quite some time since a small Qwen4 MoE might not be released.
Sources (2)
- [1]Nemotron 3.5 Bolt, Muse Spark Open, and a Fast LFM2.5 24BThe Kaitchup (Benjamin Marie) · Oct 17, 01:35 AM
- Nemotron 3.5, with a closer look at Bolt and its “native 2-bit” quantization
- Muse Spark Open: finally released, but not quite what I was hoping for
- [2]Nemotron 3.5 Bolt is okayish for small multiuser setupsr/LocalLLaMA (top, daily) · Oct 19, 02:48 AM
Extractive summary: sentences quoted from the sources.
Hype check
Independently evaluated4 independent looks since
- Evaluation · Oct 17, 2026Moss image sft (8 steps turbo lora test)
- Evaluation · Oct 17, 2026[AINews] High Taste Testers Ep. 1
- Evaluation · Oct 17, 2026T2i Moss vs. Flux3 Klein vs. Krea2
- Evaluation · Oct 17, 2026Qwen Flash Next quant and temp comparison - performance vs task time vs quality
- Follow-up · Oct 17, 2026Title: Made a ComfyUI node to fix MOSS-Image's resolution-dependent BBox coordinates
- Follow-up · Oct 17, 2026Recap on Best Open Source AI Models so far
Before this
- Oct 16, 2026Welcome Beam: a 501B open model you can own, customize, and scale
- Oct 15, 2026capy-flash: Qwen3.8-Flash-Next (125B MoE) in full NVFP4 on two 16 GB consumer GPUs
- Oct 15, 2026QUASAR 4-bit QAT checkpoints for Qwen3.5, Gemma 4 and Muse Glimmer: closer to BF16 than Google’s and NVIDIA’s 4-bit releases
- Oct 15, 2026huggingface/transformers v5.20.0: Release v5.20.0
- Oct 15, 2026NVIDIA Nemotron 3.5 Super VL Powers Specialized Agents with Multimodal Reasoning
- Oct 15, 2026Local and Distributed Large LLM Inference with MLX and RDMA on Mac Studio
