๐๐ฑ๐ด๐ฒ๐ฌ ๐๐ ๐ฟ๐ฎ๐ป๐ธ๐ #๐ฐ๐ฑ ๐ฎ๐บ๐ผ๐ป๐ด ๐๐ต๐ฒ ๐ณ๐ฎ๐๐๐ฒ๐๐-๐ด๐ฟ๐ผ๐๐ถ๐ป๐ด ๐ฒ๐บ๐ฒ๐ฟ๐ด๐ถ๐ป๐ด ๐ผ๐ฝ๐ฒ๐ป-๐๐ผ๐๐ฟ๐ฐ๐ฒ ๐ผ๐ฟ๐ด๐ฎ๐ป๐ถ๐๐ฎ๐๐ถ๐ผ๐ป๐ ๐ผ๐ป ๐๐ถ๐๐๐๐ฏ!
Put together by @commit and @Supabase, we're featured in the OSSCAR Q3 2026 Emerging ranking.
A big thank you to everyone who's explored our projects, starred our repos, or shared feedback. It's great to see our work being seen in the open-source community.
And to all the other teams on the list: hello! ๐
We're always up for conversations about efficient inference, on-device AI, and what it takes to make powerful models run on everyday hardware.
๐ฆ๐ฒ๐ฒ ๐น๐ถ๐๐: osscar.dev/org/edge0-ai
๐ง๐ฟ๐ ๐๐ฑ๐ด๐ฒ๐ฌ ๐ผ๐ป ๐๐ถ๐๐๐๐ฏ: github.com/Edge0-AI
๐๐ผ๐ถ๐ป ๐ผ๐๐ฟ ๐๐ถ๐๐ฐ๐ผ๐ฟ๐ฑ to get help with setup or share what you're building: discord.gg/Tg3W2FEuQ
๐ ๐๐๐-๐ฝ๐ฎ๐ฟ๐ฎ๐บ๐ฒ๐๐ฒ๐ฟ ๐บ๐ผ๐ฑ๐ฒ๐น, ๐ฟ๐๐ป๐ป๐ถ๐ป๐ด ๐น๐ผ๐ฐ๐ฎ๐น๐น๐ ๐ผ๐ป ๐ฎ๐ป ๐ถ๐ฃ๐ต๐ผ๐ป๐ฒ ๐๐ ๐ฃ๐ฟ๐ผ: ๐.๐ ๐๐ผ๐ธ/๐ ๐ฑ๐ฒ๐ฐ๐ผ๐ฑ๐ฒ. ๐ฑ
The obstacle: weight memory. 35B parameters (weights) ร 16 bits = ~70 GB. ๐ฆ๐๐ฒ๐ฝ ๐ ๐ผ๐ณ ๐ณ๐ถ๐๐๐ถ๐ป๐ด ๐ถ๐ ๐ผ๐ป ๐ฎ ๐ฝ๐ต๐ผ๐ป๐ฒ: ๐๐ต๐ฟ๐ถ๐ป๐ธ ๐๐ต๐ฒ ๐๐ฒ๐ถ๐ด๐ต๐๐. ๐งต
(We made it work and optimized it for iPhone 16 Pro, iOS 26.6.2.)
1/5
Step 1 done: less to store and load. But RAM is untouched. Sparsity cuts compute per token, not the bytes you have to hold, and ~19.5 GB is far more than a phoneโs RAM.
The next step is running the model without keeping that entire checkpoint in RAM.
Stay tuned for our next post on how we do it (hint: MoE)
5/5
๐ง๐ฟ๐ ๐๐ฑ๐ด๐ฒ๐ฌ ๐ผ๐ป ๐๐ถ๐๐๐๐ฏ: github.com/Edge0-AI
๐๐ผ๐ถ๐ป ๐ผ๐๐ฟ ๐๐ถ๐๐ฐ๐ผ๐ฟ๐ฑ to get help with setup or share what youโre building: discord.gg/Tg3W2FEuQ
Think your laptop's limited RAM is a barrier to your AI dreams? Think again!
๐๐ฒ๐ฟ๐ฒ ๐ฎ๐ฟ๐ฒ ๐ ๐ถ๐ป๐ฐ๐ฟ๐ฒ๐ฑ๐ถ๐ฏ๐น๐ฒ ๐ผ๐ฝ๐ฒ๐ป ๐๐ ๐บ๐ผ๐ฑ๐ฒ๐น๐ ๐๐ผ๐ ๐ฐ๐ฎ๐ป ๐ฒ๐
๐ฝ๐น๐ผ๐ฟ๐ฒ, ๐ฒ๐๐ฒ๐ป ๐๐ถ๐๐ต ๐บ๐ถ๐ป๐ถ๐บ๐ฎ๐น ๐ฟ๐ฒ๐๐ผ๐๐ฟ๐ฐ๐ฒ๐, ๐ฎ๐ ๐๐ฒ ๐๐๐ฒ๐ฝ ๐ถ๐ป๐๐ผ ๐ข๐ฐ๐๐ผ๐ฏ๐ฒ๐ฟ ๐๐๐๐.
๐๐ฒ๐โ๐ ๐ฑ๐ถ๐๐ฒ ๐ถ๐ป! ๐งต๐
๐ There's no universal winner.
RAM, latency, language coverage, and licenses all affect the choice.
๐ฃ๐ถ๐ฐ๐ธ ๐๐ต๐ฎ๐ ๐ณ๐ถ๐๐ ๐๐ผ๐๐ฟ ๐๐ผ๐ฟ๐ธ๐น๐ผ๐ฎ๐ฑ ๐ฎ๐ป๐ฑ ๐ต๐ฎ๐ฟ๐ฑ๐๐ฎ๐ฟ๐ฒ.
๐ช๐ต๐ฎ๐ ๐๐ผ๐๐น๐ฑ ๐๐ผ๐ ๐ฎ๐ฑ๐ฑ?
๐๐ถ๐ณ๐ข๐ต๐ฆ๐ฅ ๐ฃ๐บ ๐๐ฅ๐จ๐ฆ0. ๐๐ฐ๐ต ๐ข ๐ณ๐ข๐ฏ๐ฌ๐ช๐ฏ๐จ.
๐ง๐ฟ๐ ๐๐ฑ๐ด๐ฒ๐ฌ ๐ผ๐ป ๐๐ถ๐๐๐๐ฏ: github.com/Edge0-AI
๐๐ผ๐ถ๐ป ๐ผ๐๐ฟ ๐๐ถ๐๐ฐ๐ผ๐ฟ๐ฑ to get help with setup or share what youโre building: discord.gg/Tg3W2FEuQ
๐ง๐ฟ๐ ๐๐ฑ๐ด๐ฒ๐ฌ ๐ผ๐ป ๐๐ถ๐๐๐๐ฏ: github.com/Edge0-AI
๐๐ผ๐ถ๐ป ๐ผ๐๐ฟ ๐๐ถ๐๐ฐ๐ผ๐ฟ๐ฑ to get help with setup or share what youโre building: discord.gg/Tg3W2FEuQ
๐ There's no universal winner.
RAM, latency, language coverage, and licenses all affect the choice.
๐ฃ๐ถ๐ฐ๐ธ ๐๐ต๐ฎ๐ ๐ณ๐ถ๐๐ ๐๐ผ๐๐ฟ ๐๐ผ๐ฟ๐ธ๐น๐ผ๐ฎ๐ฑ ๐ฎ๐ป๐ฑ ๐ต๐ฎ๐ฟ๐ฑ๐๐ฎ๐ฟ๐ฒ.
๐ช๐ต๐ฎ๐ ๐๐ผ๐๐น๐ฑ ๐๐ผ๐ ๐ฎ๐ฑ๐ฑ?
๐๐ถ๐ณ๐ข๐ต๐ฆ๐ฅ ๐ฃ๐บ ๐๐ฅ๐จ๐ฆ0. ๐๐ฐ๐ต ๐ข ๐ณ๐ข๐ฏ๐ฌ๐ช๐ฏ๐จ.
๐ ๐ง๐ฒ๐
๐-๐๐ผ-๐ฆ๐ฝ๐ฒ๐ฒ๐ฐ๐ต (๐ง๐ง๐ฆ)
โข ๐๐๐ฑ๐ถ๐ผ๐-๐ง๐ง๐ฆ ๐.๐๐ โ 11 languages + zero-shot voice cloning huggingface.co/Edge0/Audio8-โฆ
โข Chatterbox V3 โ expressive multilingual TTS huggingface.co/ResembleAI/chโฆ
โข CosyVoice 3 โ streaming speech synthesis huggingface.co/FunAudioLLM/Fโฆ
๐๏ธ ๐ฆ๐ฝ๐ฒ๐ฒ๐ฐ๐ต ๐ฅ๐ฒ๐ฐ๐ผ๐ด๐ป๐ถ๐๐ถ๐ผ๐ป (๐๐ฆ๐ฅ)
โข ๐๐๐ฑ๐ถ๐ผ๐ด-๐๐ฆ๐ฅ-๐๐ป๐ณ๐ถ๐ป๐ถ๐๐ฒ โ continuous streaming, bounded KV cache huggingface.co/Edge0/Audio8-โฆ
โข X2Streaming-ASR โ append-only transcripts github.com/X-Square-Robot/X2โฆ
โข Qwen3-ASR-0.6B โ multilingual ASR huggingface.co/Qwen/Qwen3-ASโฆ
๐ง ๐๐ผ๐ฐ๐ฎ๐น ๐๐๐ ๐
โข ๐๐ฑ๐ด๐ฒ๐-๐๐๐ โ SSD-streamed experts to ease RAM pressure huggingface.co/Edge0/Edge0-3โฆ
โข LFM2.5-24B-A2B โ 2B active parameters huggingface.co/LiquidAI/LFM2โฆ
โข Qwen3.5-2B โ compact multimodal AI huggingface.co/Qwen/Qwen3.5-โฆ
๐ง๐ฟ๐ ๐๐ฑ๐ด๐ฒ๐ฌ ๐ผ๐ป ๐๐ถ๐๐๐๐ฏ: github.com/Edge0-AI
๐๐ผ๐ถ๐ป ๐ผ๐๐ฟ ๐๐ถ๐๐ฐ๐ผ๐ฟ๐ฑ to get help with setup or share what youโre building: discord.gg/Tg3W2FEuQ



