LHC (@StudioLHC)

@StudioLHCOct 21

The Forbidden Toys Catalog returns for a third volume. Imagine a toy that should never have been marketed and send your proposal to LHC. lhc.media/en/media/trends/op…

[Open Call] Forbidden Toys Catalogue • LHC

Vol. 3 of LHC's Forbidden Toys Catalogue: the fake Christmas catalogue turns into a product recall. Generate AI toys that don't exist, vile and flawless, as pure shelf tags (name, fake price, lying...

LHC · Oct 29, 2026 · 4:31 PM UTC

@StudioLHC3h

Four successful attacks for one dollar, but thirty times as many that ended in failure. @EnclaveAI Particle-1 occasionally proves profitable, but significantly less so as soon as it misses its target. lhc.media/en/media/ai-watch/…

An AI carries out four validated cyberattacks for one dollar, but spends thirty times more on its...

Particle-1 successfully executes four out of eleven attacks in an independent Enclave benchmark, with a cost of approximately one dollar for its validated exploits.

LHC · Oct 29, 2026 · 4:27 PM UTC

Deciding without chitchat: @Articul8_AI launches Bivia, a multimodal AI that decides on texts, images, and videos. The model returns a decision, not a monologue. lhc.media/en/media/ai-watch/…

Bivia wants to replace long AI responses with immediate and verifiable decisions • LHC

Articul8 launches Bivia, a decision-making multimodal AI capable of analyzing documents, images, and videos without generating text.

LHC · Oct 29, 2026 · 4:23 PM UTC

API compatibility, third-party models, and multi-region hosting: the stack is expanding without changing its front. @MistralAI adopts OpenAI's Responses format and hosts DeepSeek V4.1 Flash and GLM 5.3 Flash. lhc.media/en/media/ai-watch/…

Mistral adopts OpenAI's Responses format and welcomes two new competing models • LHC

Mistral launches an API compatible with OpenAI's Responses format and hosts two economical models from DeepSeek and Z.ai for AI agents.

LHC · Oct 29, 2026 · 4:20 PM UTC

Transcribing, translating, and diarizing in 40 languages: @nvidia launches Nemotron 3.5 Transcribe. The model handles up to eight speakers, completely unfazed by meeting chatter. lhc.media/en/media/ai-watch/…

Nemotron 3.5 Transcribe supports 40 languages and recognizes up to eight speakers • LHC

NVIDIA unveils Nemotron 3.5 Transcribe, an open voice model capable of transcribing 40 languages, identifying speakers, and translating meetings.

1,092

LHC · Oct 29, 2026 · 4:17 PM UTC

Image, video, voice, 3D: the creative catalog comes together under a single key. @sunra_ai announces pricing aligned with publishers', with no added margin. lhc.media/en/media/ai-watch/…

Access to creative models at publisher prices, Sunra's new offer • LHC

Sunra launches a single API to generate images, videos, voices, and 3D objects, featuring pay-as-you-go pricing and rates advertised without markups.

LHC · Oct 29, 2026 · 3:22 PM UTC

Recognizing a hesitation, a nod, or a true turn of phrase: this is the heart of the matter for voice agents. @FishAudio attempts to address this with Realtime STT. lhc.media/en/media/ai-watch/…

Fish Audio teaches voice agents to recognize the right time to respond • LHC

Fish Audio launches a real-time transcription model capable of detecting interruptions, hesitations, and speaker turns for voice agents.

LHC · Oct 29, 2026 · 3:22 PM UTC

Recognizing a hesitation, a nod, or a true turn of phrase: this is the heart of the matter for voice agents. @FishAudio attempts to address this with Realtime STT. lhc.media/en/media/ai-watch/…

Fish Audio teaches voice agents to recognize the right time to respond • LHC

Fish Audio launches a real-time transcription model capable of detecting interruptions, hesitations, and speaker turns for voice agents.

LHC · Oct 29, 2026 · 3:20 PM UTC

Shortened antibodies, scrutinized documents, and still pre-clinical trials: @AnthropicAI details what Claude actually produced, backed by figures. lhc.media/en/media/ai-watch/…

Claude enters pharmaceutical laboratories, with measured results on antibodies and experimental...

Anthropic presents initial results for Claude in pharmaceutical research, from antibody design to preclinical trials.

LHC · Oct 29, 2026 · 3:17 PM UTC

Generating text by blocks rather than token after token: that is Clover's angle. @DiffuSpace_Labs announces 680 tokens/s on a GB300, with an open source model for agents. lhc.media/en/media/ai-watch/…

DiffuSpace unveils Clover, an open-source model built for fast AI agents • LHC

DiffuSpace unveils Clover-35B-A3B, an open-source diffusion language model capable of reaching 680 tokens/s on a single NVIDIA GB300 GPU.

LHC · Oct 29, 2026 · 2:17 PM UTC

@StudioLHC6h

Video recasting leaves the labs. @eachlabs introduces Lockstep, a model that replaces characters without affecting movements or the camera. lhc.media/en/media/ai-watch/…

each::labs launches its own video character replacement model • LHC

each::labs launches Lockstep, a video recasting model capable of replacing multiple characters while preserving movements and camera angles.

LHC · Oct 29, 2026 · 2:15 PM UTC

@StudioLHC6h

Voice agents are becoming more natural, but the real question remains controlling what they say. @rimelabs launches Prism, featuring verifiable text before broadcast and fine-tuning for enterprises. lhc.media/en/media/ai-watch/…

Prism wants to make voice agents as natural as a human conversation, without losing control of...

Rime launched Prism, an S2S AI voice model with a 570 ms latency and text checkpoint, outperforming OpenAI GPT-Live-1 and Gemini 3.8 Live in τ-Voice tests.

LHC · Oct 29, 2026 · 2:11 PM UTC

@StudioLHC6h

In Claude, the legal department is getting ready-to-use tools. Citation checking, regulatory tracking, contract comparison: @AnthropicAI is strengthening its offer without making it more talkative. lhc.media/en/media/ai-watch/…

Claude transforms into a legal workspace with ready-to-use tools • LHC

Anthropic launches a legal tool gallery in Claude to analyze contracts, verify citations, and track regulatory developments.