EverPaper

[EverPaper @getEverPaper]

@NayteThaGreat 100% and it makes it so GPU and CPUs are not impacted and free for other bigger models!

[EverPaper @getEverPaper]

Releasing EverParakeet 1.1 🦜
We ported NVIDIA's Parakeet v3 to Apple's Neural Engine and fine-tuned it.
Speech to text in 25 languages, on your Mac.
Three sizes, down to 295 MB, with a lower average error than most other ports.
Core AI&ML, MLX. CC-BY-4.0. Enjoy!

https://t.co/jgS9jOgToS

[EverPaper @getEverPaper]

EverParakeet also improves WER in 22 of 25 languages.

Biggest gains: Greek (-1.51 points), Estonian (-0.99), Lithuanian (-0.92).

Italian improves too. Same Mac, same evaluation pipeline. Mean WER: 13.75 → 13.39.

[EverPaper @getEverPaper]

We’ve pushed EverParakeet Standard to 5.72% WER across 9 public English ASR benchmarks, essentially matching NVIDIA’s original Parakeet (5.58%).

Running locally on Apple’s Neural Engine, with a compact footprint.

Open ASR, no cloud required. 🦜

[EverPaper @getEverPaper]

@ClementDelangue @Google And if anyone wants to run it natively on ANE/MacOS:

[EverPaper @getEverPaper]

We ported Google's EmbeddingGemma 2 to Apple Core AI.
Text, images and audio in one embedding space, running on your Mac's Neural Engine.
Open source, Apache-2.0. Enjoy!

https://t.co/xSHEpuPKbY

[EverPaper @getEverPaper]

Concordance - a launcher for macOS that finds things by what they mean.

"the invoice for the tyres" → the PDF
"a photo of a person" → the portrait
"turn off the sound" → Toggle Mute
EmbeddingGemma 2 on the Neural Engine. Nothing leaves your Mac.

Open source ⬇️

https://t.co/zHRs4iAfaO

this is a small piece, of what EverPaper is able to do.

[EverPaper @getEverPaper]

We ported Google's EmbeddingGemma 2 to Apple Core AI.
Text, images and audio in one embedding space, running on your Mac's Neural Engine.
Open source, Apache-2.0. Enjoy!

https://t.co/xSHEpuPKbY

[EverPaper @getEverPaper]

Request an invite here, it's free! https://t.co/avwZCBD6UJ

[EverPaper @getEverPaper]

1974: we wrote our days down.
2026: we're still doing it.

Your Mac sees your whole workday.
Now it remembers it, on your Mac.
We can't wait to get your feedback on EverPaper.

Invites are limited, you will be able to invite friends once you get access yourself.

[Marco Iannello @MarcoIannello]

I wanted to remember my days, not just my meetings.

So I built Everpaper: a private daily journal for your Mac that brings together conversations, decisions, notes, and moments all searchable, all kept local.

The private beta is open and I’m looking for testers!
https://t.co/rptHSWyIzw

[Marco Iannello @MarcoIannello]

Swift, Metal and Astra are a great combo. This is the newest animation I've created for EverPaper.
It's insane how quick Astra managed to pull it off.

I will share the prompt and an opensourced (similar) animation soon on github if anyone's interested!

[Marco Iannello @MarcoIannello]

Alright, @liquidai’s new LFM2.5-2.6B Agentic model is seriously impressive on @getEverPaper . Tool calling is extremely precise, and so far it’s been noticeably better than any other consumer-ready model I’ve tested.

It’s already available to EverPaper beta testers. Not in the beta yet? Comment here, or try it with your favorite agent harness.

[Marco Iannello @MarcoIannello]

I'm building @getEverPaper, a private memory for your Mac.

A quiet, on-device record of your work: your screen, your meetings, your decisions. Nothing leaves your machine unless you say so.

Coming soon: https://t.co/nevpzofQbB
(1/4)