Sakhanda Wire
NVDA $150.96 -2.89% MSFT $508.29 +0.40% GOOGL $282.44 +0.80% META $616.38 -0.40% AMZN $267.74 +3.82%

Latest news

Older stories

MarkTechPost

Z.ai releases GLM-5.3 with post-training gains

Z.ai released GLM-5.3 on August 14, 2026, using the unchanged 743B GLM-5.2 base model. The company attributes the reported improvements entirely to expanded post-training: more long-horizon task environments, a wider range of environment types and longer training. The model is available through the Z.ai API, GLM Coding Plan and ZCode, while its weights are expected roughly two weeks after launch, following safety evaluation and hardening.

MarkTechPost

Needle 2 brings tool calling to 14MB devices

Cactus Compute has released Needle 2, an open 45M-parameter model designed for tool calling, device control and structured data extraction. The complete model is distributed as one 14MB binary and uses about 28MB of RAM for a full session. Its weights use CQ2-bit quantization, while the company’s C++ engine removes the need to install a separate runtime or download data during inference.

MarkTechPost

Tutorial outlines a compact reasoning-model training pipeline

A tutorial presents an end-to-end workflow for adapting a small language model with data from the SupraLabs reasoning corpus. Instead of downloading the full collection, the workflow streams records from the Hugging Face Hub, shuffles them, and materializes an 8,000-row sample. It examines source repositories, token lengths, task categories, and the balance between reasoning traces and final answers.

Hugging Face

Open-model growth accelerates as Qwen becomes a central ecosystem

An analysis of open models from January through August 2026 found rapid growth across public repositories. Model repositories on Hugging Face increased from 2.43 million to 2.96 million, datasets from 711,000 to 1 million, and Spaces from 1 million to 1.44 million. Activity remains concentrated: 85.6% of models have fewer than 200 lifetime downloads, while 1.5% of repositories account for 99.2% of downloads.

The Verge

Microsoft is moving Mico out of Copilot voice mode

Microsoft says its Copilot chatbot will no longer display Mico, the emotive yellow blob used in the service’s voice mode. According to a Microsoft support page, the company plans to move the avatar to Learn Live, where it will have more to react to.

TechCrunch

Writer launches Palmyra X6 to reduce AI deployment costs

Writer has introduced Palmyra X6, a flagship artificial intelligence model designed for enterprise deployments. The system is a post-training version of Z.ai’s open-source GLM-5.2. Writer says it is intended to offer deployment-ready performance at a lower per-token cost than many alternatives.

TechCrunch

Databricks raises $5 billion at a $190 billion valuation

Databricks raised $5 billion in its latest funding round, lifting the company’s valuation to $190 billion. CEO and co-founder Ali Ghodsi told TechCrunch that the company had initially planned to raise $1 billion, but investor demand expanded sharply after a report about a possible fundraising effort appeared during Databricks’ June conference. Ghodsi said selected investors expressed $15 billion in interest, prompting the company to issue more shares.

The Decoder

Suno Studio 2.0 adds chat-driven music production for Premier subscribers

Suno has released Studio 2.0, turning its AI music service into a chat-driven digital audio workstation for Premier subscribers. A beta chat function allows users to describe what they want in ordinary language and have Studio create instruments, vocals or custom plugins. Suno is not charging credits for plugin creation at present, although the company says a credit system could be introduced later.

The Verge

OpenAI chief revenue officer Denise Dresser to leave company

Denise Dresser, OpenAI’s chief revenue officer, has announced that she will leave the company in the coming weeks to pursue other opportunities. She shared the decision in a team note posted to LinkedIn. Dresser joined OpenAI in December after serving as CEO of Slack.

TechCrunch

OpenAI previews Ultrafast mode for GPT-5.6 Sol

OpenAI has introduced Ultrafast, a preview mode intended to accelerate its latest and most powerful model, GPT-5.6 Sol. The company says the mode can operate at up to 14 times the speed of standard processing and generate as many as 750 output tokens per second. Tokens are the individual pieces of text produced by a language model during an interaction.

TechCrunch

IBM partners with OpenAI to expand enterprise AI services

IBM has announced a partnership with OpenAI to bring OpenAI’s models and tools to more corporate customers through IBM’s global consulting business. The companies will jointly market AI offerings and create solutions for financial services, government, telecommunications and retail, although financial terms were not disclosed.

The Decoder

Google launches Gemini 3.7 Flash with stronger coding performance

Google has released Gemini 3.7 Flash, only three weeks after introducing Gemini 3.6 Flash. The company describes the new model as its most capable workhorse so far for coding and AI agents, attributing the rapid improvement in intelligence to algorithmic changes. The largest reported gains are in code-related tasks.