Kirill Solodskikh

White chocolate

@TheStageAI Co-founder, CEO, ex Huawei P50 AI cameras

[Kirill Solodskikh @GarchFather]

@LocallyAIApp Add the highly compressed Gemma 4 E2B and E4B! Pareto optimal! Gemma 4 E2B is 1.4 GB!

https://t.co/aot6GYjZOo

[Kirill Solodskikh @GarchFather]

@googlegemma Add these gemma4 models! Smaller, great quality! https://t.co/tWJqORvDHv

[Kirill Solodskikh @GarchFather]

@ivanfioravanti Cool! Add these smallest Gemma 4 models!

https://t.co/aot6GYjZOo

[Kirill Solodskikh @GarchFather]

@googlegemma

[Kirill Solodskikh @GarchFather]

Gemma4 E2B, compressed by @TheStageAI , from 9.3GB to 1.4GB, is running on iPhone 16e with tool calls!

The smallest and the best quality checkpoints open-sourced! @GoogleDeepMind

[Kirill Solodskikh @GarchFather]

@TheStageAI @GoogleDeepMind @Prince_Canuma, you need to add this!

[TheStage AI @TheStageAI]

The smallest checkpoints for Gemma 4 E2B and E4B for local inference. Results for E2B:

size: 9.3 GB → 1.4 GB
speed: 113 tok/s on Apple M3
quality: -3% on ifEval
runs with: MLX, llama.cpp (coming)

Pareto-optimal, open source! Links to the blog post and GitHub repo ⬇️

@GoogleDeepMind @lmstudio @ollama @huggingface @ggerganov

[TheStage AI @TheStageAI]

Try it yourself, https://t.co/6lPd1abudw

[Recraft @recraftai]

Say hello to V4.1

This model is built for images that captivate you. Photorealism is more human, gradients are dreamier, and new illustration styles are now possible.

Test it out in Recraft Studio today and see what you can create.

[Kirill Solodskikh @GarchFather]

@recraftai Very cool quality!! We are using Recraft studio!

[Kirill Solodskikh @GarchFather]

@Nick_Davidov Thank you for your permanent support, Nick!

[TheStage AI @TheStageAI]

TheStage AI Platform is now open to everyone.

Automatically accelerate your models and download them to run in the cloud or on smartphones.

[Kirill Solodskikh @GarchFather]

@TheStageAI @Beyonce , 😂

[Kirill Solodskikh @GarchFather]

@HarryStebbings @antonosika AGI, $100B investment, Safe AGI, $500B investment, Very Aligned Safe AGI, $1T investment.

If AGI becomes smarter than humans, how soon will it create something even smarter than itself?

[TheStage AI @TheStageAI]

Beyoncé heard cursing. TheWhisper heard Arsenal.

The fastest Whisper in the world.
Open-source real-time ASR.
Top 5 on OpenASR benchmarks.
1800 RTFx.

Built for live captions, transcription, and voice apps.

See the repo

[Kirill Solodskikh @GarchFather]

Self-hosted AGI starts with inference infra teams can actually run.

Well. Elastic Models v0.2.0 is much more self-serve: world’s fastest whisper-large-v3-turbo, Wan2.2 generating 5s of video in 34s on H100, and instant FLUX LoRA switching.

Explore v0.2.0

[Kirill Solodskikh @GarchFather]

Actually, comparing 1-bit with 16-bit has no sense. Everyone is using 4-bit weights with MLX. And the speed will be around 150-180 tok/s on M4 Pro. Moreover, 4-bit quantization in MLX can be done as block quantization what preserve quality for the most cases.

[PrismML @PrismML]

1-bit Bonsai 8B running locally on an M4 Pro (MLX) alongside a standard 16-bit 8B model.

Same class of model, very different deployment profile: far lower memory use and substantially higher throughput.

[Kirill Solodskikh @GarchFather]

Open-source experiments dashboard for AI researchers. Cool comparison overlays across modalities. What add next? S3 integration, authentication, model registry?

https://t.co/FMJTsq21Pf

[Kirill Solodskikh @GarchFather]

@BrianRoemmele @ClementDelangue https://t.co/rOEyzalDOX this is good