Sakhanda Wire
NVDA $155.46 -0.44% MSFT $506.24 -2.27% GOOGL $280.20 -1.24% META $618.87 -1.20% AMZN $257.89 +0.05%

MarkTechPost

397 stories · all sources

MarkTechPost

Cohere launches Embed 5 embedding models for enterprise retrieval

Cohere has released Embed 5, an embedding model family designed for enterprise search, retrieval-augmented generation and agentic retrieval. It includes Embed 5 Pro, aimed at maximum retrieval quality, and Embed 5 Fast, optimized for lower latency and cost during live queries. Both models process text, images and combined text-image inputs, support 100+ languages and accept up to 128K tokens.

MarkTechPost

NVIDIA Releases Kumo Tabular Foundation Models for Table Prediction

NVIDIA has released Kumo Tabular, a family of tabular foundation models for classification and regression. The models use labeled rows as context and predict previously unseen rows in a single forward pass, without task-specific training, hyperparameter tuning, or feature engineering. Small, Medium, and Large versions range from about 28M to 215M parameters.

MarkTechPost

Perplexity previews contextual embedding model for evidence-aware retrieval

Perplexity Research and turbopuffer have released pplx-embed-v2-context-9b-preview, a contextual embedding model designed for retrieval-augmented generation pipelines. It embeds each document chunk while considering the full document, aiming to retrieve an answer together with the material needed to verify it. The model is available as a self-hosted preview, with weights on Hugging Face under the MIT license. It is not currently offered through the Perplexity API, and its weights and interface may change without backward compatibility. Loading requires transformers>=5.4.0 and trust_remote_code=True.

MarkTechPost

Google DeepMind introduces Gemini 4 Argon with one-million-token outputs

Google DeepMind has announced Gemini 4 Argon, the first model in its Gemini 4 generation. It is designed for long-running software engineering, enterprise work in areas such as legal and finance, and cybersecurity defense. The model can produce up to 1 million tokens in one response, compared with 64K on earlier Gemini models. Google has not disclosed its input context window.

MarkTechPost

OpenAI launches GPT-6.1 Sol with lower pricing and near-Astra results

OpenAI released GPT-6.1 Sol on September 29, 2026, as an upgraded mid-tier model in its GPT-6 lineup. The hosted model is available through the OpenAI API under gpt-6.1-sol, as well as in ChatGPT Work and Codex. It is intended for agentic coding, computer use and professional tasks.

MarkTechPost

Perplexity launches Photon retrieval engine with lower production latency

Perplexity has introduced Photon, an in-house retrieval and ranking engine written in Rust. It replaces an open-source engine previously forked for the company’s AI-native search stack and now processes all production traffic. Photon also supports Fast Search in the Perplexity Search API, where clients can set search_type: "fast" on POST /search. The hosted option costs $1 per 1,000 requests; Photon itself is not open source and cannot be self-hosted.

MarkTechPost

NVIDIA team presents Physis-Lang for more physically accurate video generation

Researchers from NVIDIA, MIT and the University of Oxford have introduced Physis-Lang, a framework that uses physics-focused language to improve video world models. It adds explanations of entities, causes, interactions, governing principles, changes over time and effects to ordinary video captions. The system also creates scene-specific negative prompts describing physically implausible results, such as a stone floating on water.

MarkTechPost

RSA Introduces Agent ID to Govern Enterprise AI Agents

RSA has announced RSA Agent ID, an agentic identity security platform for regulated sectors including finance, government, healthcare and critical infrastructure. The launch took place at The AI Conference in San Francisco, where Jim Taylor, RSA’s President and Chief Product and Strategy Officer, described the security challenges created by autonomous software agents.

MarkTechPost

Liquid AI releases d1 decision model for calibrated probabilistic outputs

Liquid AI has released d1, a model designed for structured decisions rather than text generation. Given a state—plain text or a JSON object—and typed questions, it returns probabilities over predefined outcomes in one API request. The response contains zero generated output tokens. The hosted model is available through the Liquid API as d1:free; it is listed as API-only and not trainable, with no GGUF, MLX or ONNX weights for self-hosting.

MarkTechPost

OpenAI introduces persistent dots AI agents with dedicated cloud computers

OpenAI has introduced dots, persistent AI agents powered by GPT-6 Astra. Each dot receives a separate cloud computer and browser, and can continue working after the user logs off. Users assign a goal, name and boundaries, while the agent pursues the task and incorporates feedback. The first dot is included with Pro and Business Premium plans in eligible markets. Enterprise, Edu and Healthcare workspaces can access a beta when an administrator enables it. There is no self-hosted or open-weights version.

MarkTechPost

Nebius and NVIDIA Open 2026 Physical AI Awards

Nebius and NVIDIA have opened applications for the 2026 Physical AI Awards, a global competition for startups developing products in physical artificial intelligence. Applications close on October 25, 2026, while finalists and winners are expected to be announced in mid-November. There is no entry fee.

MarkTechPost

Nebius Opens 2026 Physical AI Awards for Startups

Nebius and NVIDIA have opened applications for the 2026 Physical AI Awards, a global competition for startups developing products that use artificial intelligence in the physical world. Applications close on October 25, 2026, while finalists and winners are expected to be announced in mid-November.