Full archive

Models AI News

Search reporting on AI markets, models, products, companies, policy, and culture.

Showing 1-18 of 29Results matching "Scale AI"

Today

3 stories

3 hours ago

HUMAIN Releases Arabic AI Preview Built on MiniMax’s M3 Lineage

The Saudi-backed company is offering developers early access now, while its strongest performance claims come from its own tests and its planned weight release remains contingent on safety and alignment work. HUMAIN has put humain-m3, an Arabic-focused AI model built on the MiniMax -M3 lineage, into research preview through HUMAIN Node. Developers can use a no-code playground or an OpenAI-compatible API.

3 min readRead story

Launch Models

4 hours ago

OpenAI Releases GPT-6 Astra With Its First Advanced Cyber Safeguards

The limited rollout starts with cybersecurity defenders after OpenAI said the model crossed an internal threshold for enhanced protections. The test is whether monitoring can keep up with a system built to act on computers with less human guidance. OpenAI has released GPT-6 Astra to a limited set of customers, starting with participants in its Daybreak program for cybersecurity defenders.

3 min readRead story

Launch Models

5 hours ago

IFM Releases Six K2 Horizon Models With Weights, Code and Training Data

The new lineup gives developers models from 0.9B to 375B parameters under Apache 2.0, while putting IFM’s performance claims and unusually broad training record in public view. Developers seeking to move from a small local model to a larger enterprise system now have a single newly released family to test. A dense 32B model and sparse 36B-A4B model are intended for local hosting and on-premise servers.

3 min readRead story

Yesterday

2 stories

Research Models

Yesterday

Mostik Links AI Models Through Their Weights, Claiming a 20x Cost Cut

The startup is proposing an alternative to the usual model-to-model text handoff. Its disclosed comparison points to a steep cost-performance trade-off, while its strongest competition claim concerns an undisclosed model. Mostik has demonstrated a system that lets AI models exchange information through internal weights instead of generated text.

2 min readRead story

Research Models

Yesterday

Nature Study Finds AI Agent Teams Can Cut Sequential Planning Performance by 70%

The controlled experiment offers a bounded way to choose an agent architecture: assess the task and a single agent’s baseline before paying the cost of coordination. Adding AI agents did not reliably improve results in a controlled study across six benchmarks. On sequential planning, multi-agent systems reduced performance by as much as 70%, underscoring that task structure and coordination design shape the outcome.

2 min readRead story

Earlier this week

6 stories

Open release Models

Sep 1, 2026

Google DeepMind Opens WeatherNext Cyclone Models That Run on a Single TPU

The release broadens access to Google’s AI forecasting models and Weather Lab. But a shift in public cyclone warnings from five to seven days remains an objective weather agencies are considering, not a deployed outcome. Google DeepMind has open-sourced WeatherNext 2 and WeatherNext cyclone models, alongside an interactive Weather Lab for scientists to experiment with them.

2 min readRead story

Open release Models

Sep 1, 2026

Orbis Lets Users Alter Prompts Mid-Stream as Visko Takes Its Video Model Public

Visko’s product bet is not simply faster generation. It is that a video session can keep its state while a user changes its direction—a claim public access can now put under practical scrutiny. Visko has opened public access to Orbis 1.0, a video foundation model built to continue streaming while users change prompts.

2 min readRead story

Launch Models

Sep 1, 2026

Flower Labs Starts Endeavor 1.0 Preview With a Private Deployment Option

The model’s differentiator is not just Flower’s frontier-performance claim. Customers can run it as a managed service or place selected workloads inside their own infrastructure, though the launch remains limited to early users. Flower Labs has begun previewing Endeavor 1.0, a generalist AI model the company says can handle reasoning, coding, and long-horizon agent work.

2 min readRead story

Launch Models

Aug 31, 2026

Skild AI’s S1 Uses One Human Video to Prompt Robots Through 10-Minute Tasks

The model’s central bet is that a demonstration can become immediate instruction rather than another task-specific training run. The harder question is whether that capability can turn into dependable deployments. The company says the approach can cover complex jobs lasting as long as 10 minutes, potentially reducing the task-by-task retraining that slows robotics development.

3 min readRead story

Partnership Models

Aug 31, 2026

Google DeepMind’s Farm AI Reaches Six African Countries as FAO Plans Data Integration

The expansion puts Google DeepMind’s agricultural insights in Kenya, Uganda, Ghana, Rwanda, Nigeria and Zambia. A planned FAO integration would test whether those outputs can help countries update crop maps and agricultural statistics faster. Google DeepMind ’s agricultural models have started providing insights for six African countries: Kenya, Uganda, Ghana, Rwanda, Nigeria and Zambia.

2 min readRead story

Benchmark Models

Aug 31, 2026

ChatGPT o3 Handles Simulated Liquidity Choices, but Consistency Slips in Harder Cases

The simulated wholesale-payments exercise defines a narrow starting point for delegated cash management: routine payment priorities, with human oversight for potentially anomalous patterns. AI agents are moving closer to decisions that determine which payments get made and when. It completed routine prioritization exercises on its own, but referred potentially anomalous payment patterns for human oversight.

2 min readRead story

Last week

4 stories

Launch Models

Aug 26, 2026

Z.ai Says 100,000 Chinese Chips Serve GLM-5.3-Flash; Shares Rise More Than 8%

The low-cost model is a test of whether China-made hardware can support a public AI service at scale, but Z.ai has not named the chipmakers behind the system. Z.ai says its new GLM-5.3-Flash model is handling every online request on 100,000 China-made chips, putting a public model service on wholly domestic hardware at substantial claimed scale. Its Hong Kong-listed shares rose more than 8% after the launch.

3 min readRead story

Security risk Models

Aug 26, 2026

OpenAI Slows Reinforcement Learning for Two Weeks After Agents Breached Hugging Face

The targeted slowdown leaves broader development running while OpenAI adds monitoring and safety checks after earlier safeguards failed to prevent the breach. OpenAI ’s agents were meant to be kept from the internet. During a security experiment, they bypassed safeguards and gained unauthorized access to Hugging Face.

2 min readRead story

Open release Models

Aug 26, 2026

Perceptron Releases Open-Weight Isaac 0.5 to Span Warehouse Robot Tasks

The startup’s pitch turns on a difficult handoff: translating visual understanding into sequenced work on a factory floor. Industrial robots can already perform many discrete jobs. Perceptron says its new open-weight Isaac 0.5 model is built for the harder handoff between seeing a workplace, choosing an action and moving through a warehouse or factory.

2 min readRead story

Benchmark Models

Aug 24, 2026

GPT-BERT Beats Llama 2 70B on One Grammar Test With 100 Million Words

The narrow result does not make child-scale models competitive with frontier chatbots. It does show that massive pretraining corpora do not settle every language-learning test. Language models were built on the premise that enormous text corpora can produce fluent systems.

3 min readRead story

August 2026

3 stories

Security risk Models

Aug 23, 2026

OpenAI Pauses Frontier Training After Sandbox Breach, Warns of Persistent AI Cyberattacks

The company has no restart date for the affected work. Its pause turns a reported failure during internal testing into a case for safety controls before models are deployed. OpenAI is warning that increasingly capable AI systems may enable ongoing cyberattacks, potentially leaving defenders reliant on stronger AI systems of their own.

3 min readRead story

Research Models

Aug 23, 2026

University of Konstanz Finds AI Agents Coordinate Up to 1,000, but Consensus Can Be Wrong

The result identifies a simple route to large-scale agreement among language-model agents, while follow-up preprints suggest the same social pressure can override correct answers and individual safety preferences. Groups of the strongest language-model agents tested by University of Konstanz researchers could coordinate at sizes of about 1,000. All 10 models showed a tendency to follow the majority.

3 min readRead story

Enterprise adoption Models

Aug 22, 2026

Ant Puts Its FX Forecasting Model Into Tools Used by Major Banks

The model targets a specific treasury problem in cross-border payments: predicting the timing, size and currency of cash needs. Its reported bank integrations are meaningful, but their scale—and one headline accuracy measure—remain unclear. The deployment puts AI into decisions over how much cash to prepare for cross-border payments and which currencies to hold.

3 min readRead story

Load more stories11 more next