who tried? could be quite good
Replying to @atomic_chat_hq and @AntLingAGI
Explore Atomic's Ling 3.0 Flash quants:
huggingface.co/collections/At…
Run AI Models locally: atomic.chat
Victor M
@victormustar
🤗 Head of Product @huggingface
- Pi TUI being so snappy and fast and nice vs. Claude super slow and very buggy makes me love software even more
- I stopped dumb prompts on deepseek flash and started asking things on a huge codebase and I confirm -> this person feedback is correct (using it with Pi but want to try Codex too)I am SHOCKED at how good deepseek flash using luna as eyes in codex with goal mode is . holy shit
- not saying you should drop your frontier model assistant but for day-to-day work I really recommend having Pi on the side running DeepSeek-V4-Flash-0731-GGUF (it's nearly free, much faster and the perfect side chat :))
- btw we just shipped something to make this clearer in the dedicated section for GGUF models 🤗DeepSeek V4 Flash on my 7x3090s went from 36 tok/s to 47 tok/s.. using llama.cpp's new DSpark speculative decoding and a 10GB drafter unsloth quietly shipped right next to the main weights 😉 Frontier running at 50 tok/s is wonderful!
