Nativ brings LiquidAI LFM2.5 to Mac locally: 82 tok/s decode, <8.5GB RAM for 128K context
JosephJacks_ · x · 2026-08-06
Open-source local AI tool Nativ has announced Day 0 support for the newly released LFM2.5-2.6B model by LiquidAI.
Running on Apple Silicon Macs, the model operates without quantization (Full BF16) and delivers impressive performance:
- Prefill speed: 11,000+ tokens/s
- Decode speed: 82 tokens/s
- Memory usage: Peak memory under 8.5GB (even at full 128K context)
- Concurrency: Scales to 476 tokens/s aggregate across 16 concurrent requests
Nativ is a free, open-source macOS app optimized via MLX-VLM, supporting text, vision, video, code, and audio. It focuses on a fully local, account-free, and subscription-free deployment experience.
Related event: LiquidAI LFM2.5 Runs Locally on Mac with Strong Performance(3 posts)→
More from Infra
- OpenAI removes Ultrafast tier from GPT-5.6 Sol in Codex, fueling GPT-6 Sol rumors — imjustnewatai · 2026-09-22
- Is upgrading from 2x to 4x RTX 3090 worth it for local LLM work? — fgoricha · 2026-09-22
- Qwen-Image local on a 24GB MacBook Pro takes 5-6 minutes per image — vista8 · 2026-09-22
- Cerebras CEO on Jensen Huang: a decade trading as 'nobody' before Nvidia made it — rohanpaul_ai · 2026-09-22
- How Tencent Hunyuan packed a 770B model into 214 GiB with 5-bit-per-4-weights quantization — TencentHunyuan · 2026-09-22
- 456GB DeepSeek v4.1 runs locally at 40 tok/s with Threadripper + dual RTX 6000 hybrid setup — HankYeomans · 2026-09-22