19+ hour ago   (475+ words) The AllSpark team has released Iris-mini and Iris-pro, two open-source search agents along with a full training recipe. According to the paper, the training data and models also improved performance on tasks they were never trained for, including general tool…...

3+ day, 19+ hour ago   (509+ words) Deepseek's new multimodal model is built mainly to cut the operating costs of long contexts. The biggest gain is in memory use, though Deepseek also promises better model performance. According to the technical report, Deepseek has a clear goal with…...

1+ week, 22+ hour ago   (483+ words) Meta's Superintelligence Labs have released their first real-time audio perception model. It transcribes speech, tells speakers apart, and detects sentence boundaries during live conversation. The Spark-family model breaks incoming audio into 80-millisecond chunks. After each one, it decides whether to…...

1+ week, 23+ hour ago   (647+ words) Abliteration.ai removes trained refusal mechanisms from powerful open-weight models and sells access to the modified versions as a service. There's a legitimate market for that, but the same setup creates a difficult security trade-off. Anyone with access to an…...

1+ week, 4+ day ago   (289+ words) Three weeks after Gemini 3.7 Flash, Google is back with Gemini 3.8 Flash. The budget model is supposed to beat its predecessor by a wide margin in coding. According to Google's own numbers, Gemini 3.8 Flash hits 73.7 percent on the DeepSWE v1.1 benchmark for…...

1+ week, 5+ day ago   (392+ words) AI company Runway has unveiled Solaris, the first model in a new category it calls "Interface World Models." Instead of running code, the system generates the user interface frame by frame as you interact with it. Runway lists several use…...

2+ week, 1+ day ago   (544+ words) Researchers at Google Research have introduced WikiSkill, a framework that pairs AI agents with a persistent knowledge base. Instead of throwing away what the agent learned after each run, the system collects knowledge about failures and successes in a wiki-like…...

2+ week, 4+ day ago   (419+ words) Alibaba's Qwen team is introducing Qwen3.8-Flash-Next, a multimodal mixture-of-experts model that serves as an architecture preview of Qwen4. It aims to match much larger models at a fraction of the training cost. The model has 125 billion total parameters but only activates…...

2+ week, 4+ day ago   (212+ words) the-decoder.com IBM drops open-weight Granite 4.2 family with built-in agentic capabilities under Apache 2.0 IBM has released its Granite 4.2 language models in 3B, 8B, and 30B sizes. Trained from scratch on about 15 trillion tokens, the models support context windows up to 512,000 tokens and can…...

3+ week, 1+ day ago   (302+ words) Skills are seen as a practical way to make AI agents more capable without retraining them. A new study shows why they work and where they fall short. A team of researchers from Princeton University, UC San Diego, and other…...