/Tech16h ago

Over 25 open-weight models launch in a single week, including NVIDIA's 550B Nemotron and Google's Gemma 4 12B

Releases include Ideogram's first open 9.3B image generation model.

1132.4K3292.5K435.2K

#19

Original post

Victor M@victormustar

Before the week ends, let's acknowledge one of the most INSANE week ever for open AI, with 25+ notable open-weight drops across every modality:

🧠 LLMs

→ NVIDIA Nemotron 3 Ultra: 550B hybrid Mamba-MoE, only 55B active, 1M context, MMLU 89.1. NVFP4 variant claims ~5x throughput on Blackwell. First openly-weighted 550B hybrid Mamba-Transformer, closing the gap with frontier closed models.

→ Google Gemma 4 12B: fully open dense any-to-any (text/image/audio/video), 256k context, encoder-free, 140+ languages, AIME 2026 at 77.5. Shipped with a 23-checkpoint QAT wave (mobile ONNX + MLX). Most deployable model of the week.

→ StepFun Step-3.7-Flash: 198B sparse MoE VLM, ~11B active, SWE-Bench PRO 56.3. Apache 2.0.

→ Liquid AI LFM2.5-8B-A1B: edge MoE, just 1.5B active, 128k ctx, MATH500 88.8, MLX-ready. Best on-device option this week.

→ JetBrains Mellum2-12B-A2.5B-Thinking: their first open MoE, near-Qwen3-14B coding at 2.5B active. Apache 2.0.

🎨 Image gen (the surprise of the week)

→ Ideogram 4: their FIRST-EVER open weights. 9.3B flow-matching DiT trained from scratch. #2 overall behind GPT Image 2, top open-weight model on Design Arena + LMArena. Strongest open checkpoint for text-rich images, full stop. It has taste. Still can't believe this is open weights.

🔊 Audio & Speech (a breakout week for open TTS, 4 labs shipped)

→ Boson Higgs Audio v3 4B: 102 languages, 21 emotions, singing/whispering/shouting, sub-second TTFA. → RedNote dots.tts: the only fully continuous (no codec) open TTS pipeline, Apache 2.0. → Google Magenta RealTime 2: real-time music gen, <200ms latency, text+audio+MIDI. multimodalart ported it to PyTorch within hours with live ZeroGPU demos. → NVIDIA Nemotron-3.5 ASR: 600M streaming, 17x more concurrent streams vs Parakeet RNNT 1.1B.

👁️ Vision & VLMs

→ PaddleOCR-VL-1.6: SOTA document parsing at 1B params, Apache 2.0. → Baidu NAVA: 6.3B joint audio-video gen, best-in-class A/V sync, Apache 2.0.

🎬 Video, 3D & World Models

→ NVIDIA Cosmos3-Super: 64B omnimodal world model coupling action trajectories with video+audio gen, for Physical AI. → JD JoyAI-Echo: up to 5-min multi-shot text-to-video on LTX-2.3. → ByteDance Bernini-R + VAST TripoSplat (single-image-to-3D Gaussian splats, MIT).

2:59 PM · Jun 5, 2026 · 435.2K Views

/Tech16h ago

Over 25 open-weight models launch in a single week, including NVIDIA's 550B Nemotron and Google's Gemma 4 12B

Releases include Ideogram's first open 9.3B image generation model.

1132.4K3292.5K435.2K

#19

Original post

Victor M@victormustar

Before the week ends, let's acknowledge one of the most INSANE week ever for open AI, with 25+ notable open-weight drops across every modality:

🧠 LLMs

→ StepFun Step-3.7-Flash: 198B sparse MoE VLM, ~11B active, SWE-Bench PRO 56.3. Apache 2.0.

→ Liquid AI LFM2.5-8B-A1B: edge MoE, just 1.5B active, 128k ctx, MATH500 88.8, MLX-ready. Best on-device option this week.

→ JetBrains Mellum2-12B-A2.5B-Thinking: their first open MoE, near-Qwen3-14B coding at 2.5B active. Apache 2.0.

🎨 Image gen (the surprise of the week)

🔊 Audio & Speech (a breakout week for open TTS, 4 labs shipped)

👁️ Vision & VLMs

→ PaddleOCR-VL-1.6: SOTA document parsing at 1B params, Apache 2.0. → Baidu NAVA: 6.3B joint audio-video gen, best-in-class A/V sync, Apache 2.0.

🎬 Video, 3D & World Models

2:59 PM · Jun 5, 2026 · 435.2K Views

Sentiment

Users are excited about the surge in open-weight AI model releases across modalities due to their volume, quality, rapid progress, and new capabilities like 1M context, while a few dismiss them as insignificant or unscalable.

Pos

90.5%

Neg

9.5%

23 comments with sentiment.

Cluster Engagement

Posts from X

Most Activity

VIEWS6.3KBOOKMARKS30LIKES21

Victor M@victormustar

Let’s not forget:

https://huggingface.co/Hcompany/Holo-3.1-4B

And from yesterday

https://huggingface.co/collections/google/gemma-4-qat-q4-0

1d6.3K2130

RETWEETS329

Victor M@victormustar

Before the week ends, let's acknowledge one of the most INSANE week ever for open AI, with 25+ notable open-weight drops across every modality:

🧠 LLMs

→ StepFun Step-3.7-Flash: 198B sparse MoE VLM, ~11B active, SWE-Bench PRO 56.3. Apache 2.0.

→ Liquid AI LFM2.5-8B-A1B: edge MoE, just 1.5B active, 128k ctx, MATH500 88.8, MLX-ready. Best on-device option this week.

→ JetBrains Mellum2-12B-A2.5B-Thinking: their first open MoE, near-Qwen3-14B coding at 2.5B active. Apache 2.0.

🎨 Image gen (the surprise of the week)

🔊 Audio & Speech (a breakout week for open TTS, 4 labs shipped)

👁️ Vision & VLMs

→ PaddleOCR-VL-1.6: SOTA document parsing at 1B params, Apache 2.0. → Baidu NAVA: 6.3B joint audio-video gen, best-in-class A/V sync, Apache 2.0.

🎬 Video, 3D & World Models

1d435.2K2.4K2.5K

REPLIES2

Levi Fawcett@Levi_Fawcett

@victormustar We can't manually evaluate every model at this point.. is there someone who evals/benchmarks each model, with examples? Would follow

1d2.8K52

Victor M@victormustar

@Levi_Fawcett your agents will 🚀 against your real project

1d2.2K134

Milan@influenist

@victormustar The model race is becoming a commodity race.

The next winners won't be determined by model names, but by products, agents and distribution.

1d1.8K63

NeoView@view_neo58565

@victormustar been playing with Bernini and the editing is legit. https://github.com/bytedance/Bernini

4h5532

Virgil Maro@_virgil19

@victormustar 25 frontier drops in one week means the weights are the commodity now. whatever's scarce moved up a layer, to the harness and evals that make one actually usable

1d1.5K51

Kenneth Hurley@SuperGeniusEth

@victormustar So pumped about the direction of all this, as it really is what @elonmusk envisioned for truly democratized AI and, IMHO, will end up crushing OpenAI.

22h6805

CJ@cjtrade01

@victormustar People sleeping on Qwen 3.7. Best OS LLM.

1d49721

BOB CHEN@Bobchenjingbo

@victormustar the open-weight cadence is the real story of the year. a year ago you waited months for a frontier-ish model; now it's 25 in a week.

the bottleneck moved from "can i get a good model" to "can i evaluate them fast enough to choose." curation is the new scarcity.

14h32821

Richard Palethorpe@jichiep

@victormustar It's easy to underestimate the importance of the smaller task specific models

1d1K5

Howdy 🎒@HadiHNCA

@victormustar @julien_c Use them free at scale by http://flockllm.com It’s open source.

1d2251

Mary Newhauser@m_newhaus

@victormustar @NielsRogge Literally the worst week to be on vacation. Maximum fomo 🥲

1d9742

Alex@Myrlikte

@victormustar idiogram lol

1d2702

Jim@JimCRowan

@victormustar @pmarca Open source is not getting enough attention in the enterprise… we all want to talk token maxing but let’s talk token value, that happens with these open source models doing real work

19h2531