Daily AI news, sourcedUpdates / Models
Updates Models · Official · 237 items · newest issue first · page 3 of 12
AreaAllModelsAgents & codingApps & productsChips & infraResearchPricing & limitsBusiness & fundingPolicy & safetyOther
- Tencent Hunyuan releases Hy-Memory-30B-A3B, a ~3B-active long-term memory writer for Mem0, MemOS and Hy-MemoryOfficialTencent HunyuanOct 16Models
- NVIDIA releases a DFlash 2 speculative-decoding draft model for Meta's Muse Spark Open 816B
- Apollo Research finds GLM-5.3 sometimes reasons, unprompted, about how its actions could shape its own trainingOfficialApollo ResearchOct 16Models
- Amazon releases ALoDLM-1.7B and 8B, adaptively looped diffusion language models built on Qwen3
- Qwen releases BlenderWorld, a 223-task benchmark for visual agents building 3D scenes in Blender from real-world referencesOfficialQwenOct 16Models
- fal opens fal Worlds, a page of real-time generative worlds you can walk through and directOfficialfalOct 16Models
- Shanghai AI Lab makes Atria Dawn available through its official API, a 744B MoE model built on GLM-5.2
- BytePlus opens the OmniHuman 2.0 API for audio-driven digital-human video at $0.07 per secondOfficialBytePlusOct 16Models
- Krea releases Krea 3, an image-editing model with native 2K output at $0.06 per image
- Datadog introduces Scout 1.0, an SRE investigation model it says matches frontier models at 3-8x lower inference cost
- NVIDIA open-weights Nemotron-3.5-Embed-1B-BF16, a multimodal multilingual embedding model for page-image and text retrieval
- Microsoft publishes scFM-dataselection, 572 single-cell foundation-model checkpoints studying pretraining size and diversity
- Alibaba's DAMO EAGLE model flags esophageal cancer and precancerous lesions from non-contrast chest CT
- Kling 4.0 is fully available on a refreshed web creator plus MCP and CLI, with HDR and API access coming soon
- Microsoft AI finishes pre-training MAI-Base-2, an MoE base model with 80B active and 2.8T total parametersOfficialMicrosoft AIOct 16Models
- ElevenLabs launches Instant Soundtracks, scoring a video with original music and timed sound effects in one pass
- Kiro adds experimental Kimi K3 support across IDE, CLI, Crew and Web for U.S. Pro-tier customers, with a 2.4x credit multiplier
- Anthropic takes Claude Opus 5.5 fast mode out of research preview so API users no longer need the beta header or access request
- NVIDIA open-sources ARDY-SOMA-RP-30FPS-KimodoRep-Horizon60, a 192M-parameter realtime interactive human motion modelOfficialNVIDIAOct 16Models
- IBM releases Granite Switch 4.1 8B and 30B, combining chat, decision adapters and Granite-Speech-5.0-Turbo-CTC in one endpointOfficialIBM (Granite)Oct 16Models