Connect with us

Technology

AutoTrust AI Releases JEV-27B, an Open Decision Model for Self-Hosted AI Agents

Published

on

Trained in about 9.2 hours on one NVIDIA B200, JEV-27B adds fast, calibrated System 1 decisions to a frozen Qwen3.8-27B backbone while preserving its System 2 generation path.

SINGAPORE, Sept. 29, 2026 /PRNewswire/ — AutoTrust AI today released JEV-27B, an Apache-2.0 open-weights model designed to handle frequent, structured decisions inside AI agent workflows while retaining the underlying model’s full generation and reasoning path.

JEV-27B answers yes/no, multiple-choice and 0–5 rating questions in a single forward pass and returns a calibrated probability for every option. It runs on one NVIDIA B200 inside a customer’s own infrastructure and serves both fast System 1 decisions and deliberate System 2 generation from a single set of weights.

The model trains a 108.9-million-parameter decision block—about 0.4% of the full model—on top of a frozen Qwen3.8-27B backbone. AutoTrust reports that training took approximately 9.2 B200-hours. With the decision block switched off, all 164 HumanEval completions were byte-identical to those produced by the base model.

“Jev proved there is real demand for models that decide rather than write,” said Daniel Tang, AutoTrust AI’s chief executive and co-founder. “JEV-27B shows that this capability can run on one GPU inside a customer’s own infrastructure, next to a reasoning model. For companies that cannot send every decision to a third-party API, that changes both the cost and the risk.”

Evaluation and evidence

AutoTrust AI evaluated JEV-27B across six public text-decision benchmark groups. It reported scores of 88.70% on JevBench, 83.75% on Kev, 73.89% on OpenJev text, 92.91% on Nimble, 77.46% on VitaminC and 87.71% on MASSIVE-en, for an equal-weight six-group mean of 84.07%.

For additional context, AutoTrust AI also ran the hosted TypeSafe Jev 1.13 API on the same benchmark groups and reported a six-group mean of 83.85%, with JEV-27B scoring higher on four groups and lower on two. Because AutoTrust conducted this comparison itself, the figures should be read as internal comparative evidence—not as an independent third-party validation or a claim of across-the-board superiority.

For public baseline context, AutoTrust reproduced the scores published by TokenRhythm for NeoHorse-Jev, Open-Jev, Kev and Laya English. AutoTrust did not rerun those four external baselines. The pinned source table is available at https://huggingface.co/TokenRhythm/NeoHorse-Jev-4B/blob/b50e043e22e0e41e7fc0c244e4daa707b8124930/README.md. These figures are included as published benchmark context rather than as a new same-environment comparison by AutoTrust.

JEV-27B was also evaluated for fidelity to its distillation target. On 25,376 held-out questions labeled with Jev 1.13 probability distributions, JEV-27B reported a mean KL divergence of 0.017, where zero means identical distributions. On decision-models-under-pressure, an independent benchmark scored against human labels, JEV-27B reached 96% of Jev 1.13’s accuracy with 16 answer options. In that independent test, JEV-27B approached—but did not exceed—Jev 1.13.

AutoTrust AI measured a median decision latency of 137 milliseconds and sustained throughput of about 130 decisions per second on one B200. The model card also cites third-party measurements of 238 to 301 milliseconds and 23 decisions per second for Jev’s hosted API. These are not controlled, like-for-like results: the hosted API measurements include network time, while AutoTrust AI’s local measurements do not, and the hardware, serving and concurrency conditions differ.

How it works

AutoTrust AI uses the terms System 1 for fast, typed decisions and System 2 for deliberate generation and reasoning. JEV-27B serves both from one set of weights. It follows JEV-9B as the company’s second integrated System 1 and System 2 open model.

The model is built with AutoTrust AI’s Blocks of Experts recipe. A strong pretrained model, Alibaba’s open-weights Qwen3.8-27B, stays frozen as one expert block. A small, detachable block is trained for a single skill, and a router sends each request either to the fast decision block or to the deliberate generation block.

The decision block holds 108.9 million trained parameters, 0.4% of the model, and took about 9.2 hours to train on one NVIDIA B200. The reasoning path was left untouched. With the decision block switched off, JEV-27B scores 78.0% on the HumanEval coding test, and all 164 of its completions are byte-identical to the base model’s.

In a demonstration reel released with the model, a self-hosted JEV-27B served as the decision engine for 10 tasks. It played Doom, making 64 decisions in a target-practice scenario, and steered a simulated drone through a MuJoCo obstacle course. It ran a live Google Flights search from Zurich to London and verified 21 results, navigated Wikipedia to Gödel’s incompleteness theorems, flagged four regression risks in a sample change to authorization code and routed a billing-refund ticket to support.

“Every AI agent is really a long chain of small decisions—which button to press, which file to open, which queue a ticket belongs in,” said Josh Liu, AutoTrust AI’s chairman and co-founder. “Make each one fast, private and cheap, and you change the economics of the whole chain.”

AutoTrust AI said JEV-27B inherits Jev 1.13’s blind spots, including multi-hop reasoning, arithmetic, dates and adversarial inputs, and that its training data is English-centric. The company says the model is not meant for high-stakes decisions and recommends gating its answers on confidence.

Benchmark context and source notes

Scores are in percent. JEV-27B and the hosted TypeSafe Jev 1.13 API were measured by AutoTrust AI; this is not third-party validation. NeoHorse-Jev, Open-Jev, Kev and Laya English are published baselines reproduced from TokenRhythm and were not rerun by AutoTrust. TokenRhythm source (pinned revision): https://huggingface.co/TokenRhythm/NeoHorse-Jev-4B/blob/b50e043e22e0e41e7fc0c244e4daa707b8124930/README.md. Full AutoTrust methodology and source notes: https://huggingface.co/autotrust/JEV-27B.

Availability

JEV-27B is available under the Apache-2.0 license at huggingface.co/autotrust/JEV-27B. The release includes the weights, decision adapter, training and serving code, vLLM support and full evaluation reports. A demonstration reel is available at https://huggingface.co/spaces/autotrust/JEV-27B-Demo.

AutoTrust AI plans to build the JEV decision block into future models in its Guru family, which powers the ScienceGuru research platform. ScienceGuru is available for Windows and macOS at https://scienceguru.ai/. AutoTrust AI also offers customized sovereign deployments for enterprises.

JEV-27B was trained on SargeDev/jev-distill-corpus-v3, a public, Apache-2.0-licensed corpus of Jev 1.13’s outputs. It shares no weights or code with, and is not affiliated with or endorsed by, TypeSafe AI. Jev and TypeSafe are trademarks of their respective owners.

About AutoTrust AI

AutoTrust AI Pte. Ltd. is a Singapore-incorporated AI research company building the Guru family of foundation models and ScienceGuru, an AI research platform for scientists and research teams. Its Blocks of Experts architecture combines pretrained expert blocks with small trained adapters to build frontier-capable and sovereign models efficiently. Learn more at autotrust.ai.

View original content to download multimedia:https://www.prnewswire.com/apac/news-releases/autotrust-ai-releases-jev-27b-an-open-decision-model-for-self-hosted-ai-agents-302891725.html

SOURCE AutoTrust AI

Continue Reading

Technology

AiID Emerges from Stealth with Frictionless ID™

Published

on

By

Fake IDs can look perfect. Real IDs shouldn’t have to.

WEST PALM BEACH, Fla., Oct. 13, 2026 /PRNewswire/ — After two years in stealth, AiID today unveiled Frictionless ID™, a credential intelligence platform built to improve a critical first step in identity verification: determining whether the ID itself is genuine.

Now operating at beta locations, AiID is taking on both sides of a costly problem: counterfeit credentials that get through and genuine IDs that get turned away.

In three product demonstration examples, enabling AiID’s intelligence layer changed a genuine ID from FAIL to PASS and two counterfeit IDs from PASS to FAIL.

“When a system repeatedly rejects genuine IDs, it does more than frustrate customers. It undermines confidence in the result,” said Brandon Currul, founder and CEO of AiID. “We built Frictionless ID™ to restore that confidence, with a clear objective: catch counterfeit credentials and stop unnecessarily rejecting genuine ones. Stop asking a good ID to prove it’s perfect.”

AiID has developed a proprietary intelligence layer that moves beyond how closely an ID resembles an ideal sample to evaluate the authenticity of the credential itself. Before facial matching, liveness or broader identity verification, its engine applies validated document intelligence to answer the question that comes first: Is this ID genuine?

The company’s commercialization strategy includes direct integrations, embedded licensing for identity and fraud-prevention platforms, and complete solutions combining credential intelligence with capture, biometrics and optional compliance services. The platform is designed to strengthen the systems businesses already use, not force them to start over.

AiID has demonstrated phone-image, API and end-to-end workflows. The company is now focused on scaling Frictionless ID™ for broader commercial availability.

AiID has begun investor discussions for its seed round. During Venture Atlanta week, the company will advance those conversations and explore commercial and integration partnerships to bring Frictionless ID™ to more businesses and verification platforms.

To arrange a demonstration or discuss investment and partnership opportunities, visit aiid.io/connect.

About AiID

AiID develops Frictionless ID™, proprietary credential intelligence built to strengthen document authentication and reduce friction across identity-verification workflows.

Media Contact

Brandon Currul, Founder & CEO
424595@email4pr.com | 508-321-1671

View original content to download multimedia:https://www.prnewswire.com/news-releases/aiid-emerges-from-stealth-with-frictionless-id-302906521.html

SOURCE AiID

Continue Reading

Technology

Building the Future of Montessori Technology: How The Montessori Foundation and illumine Are Shaping What Comes Next

Published

on

By

WILMINGTON, Del., Oct. 13, 2026 /PRNewswire/ — For more than a century, Montessori education has been built around careful observation, independence, purposeful environments and deep respect for the individual child. The technology supporting Montessori schools, however, has often evolved along a very different path.

Many schools still operate across a patchwork of systems for admissions, enrollment, billing, attendance, parent communication, observations, records and school management. As organizations expand across multiple campuses, the result can be duplicated work, disconnected information and increasing complexity for already lean teams.

Now, that is beginning to change.

An ongoing collaboration between illumine and The Montessori Foundation, led by President Tim Seldin, is bringing Montessori expertise and technology together to rethink how schools can operate in a more connected, increasingly AI-powered world.

The shift is already translating into adoption. Over the past six months, more than 100 Montessori schools have begun using illumine, with schools across California, New York, Texas, Florida and other parts of the United States moving away from fragmented technology stacks. The objective is not simply to bring more technology into Montessori schools. It is to build technology that understands them.

When Montessori expertise met technology

The collaboration began at an industry event, where Seldin encountered illumine’s childcare and preschool management software firsthand.

What caught his attention was not simply the breadth of the platform, but its potential to address some of the operational realities facing Montessori organizations. That first interaction led to a deeper conversation, with Seldin helping the illumine team understand Montessori not simply as an educational philosophy, but as an operating environment with its own structures, workflows and priorities.

For more than three decades, The Montessori Foundation has worked with Montessori schools around the world to help them become stronger, more sustainable and more effective. That experience has given the Foundation a deep understanding of the realities facing Montessori school leaders, educators and families.

Through ongoing conversations with Seldin, illumine gained a closer understanding of how Montessori schools manage admissions, enrollment, family communication, observations and day-to-day operation, and how frequently those processes span multiple disconnected systems.

For illumine, the lesson was not that Montessori schools are resistant to technology. It was almost the opposite. Montessori leaders are looking for technology that understands their work. Together, the organizations are demonstrating what becomes possible when deep Montessori expertise and technology expertise inform one another from the outset.

From collaboration to impact

For schools that have already moved to illumine, the shift is about more than replacing individual software tools. Connecting previously separate systems can reduce the cost and complexity of managing multiple platforms while creating a more consistent experience for guides, administrators and families.

“We’re seeing Montessori schools across the country rethink how they manage their daily operations,” said Navneet Rastogi, Founder and CEO of illumine. “We first saw this shift take shape in California, where our initial wave of early adopters emerged in 2025. The momentum accelerated significantly this year with the launch of our purpose-built Montessori record-keeping tools for student observations and lesson planning within illumine.

Today, we’re seeing this trend extend beyond California to states like New York, Texas, and Florida, as more Montessori schools move away from fragmented systems and bring their administrative and academic record-keeping operations into one connected platform. This shift is helping schools reduce the cost and complexity of managing multiple tools while creating a more seamless experience for educators, administrators, and parents.

We’re continuing to invest in this technology because we believe Montessori schools deserve systems that not only support their day-to-day operations but also evolve alongside the ambitions of the educators and organizations using them.”

The collaboration is also expanding the conversation around what technology can mean for Montessori. Through webinars, live workshops and conversations with Montessori experts, illumine and The Montessori Foundation have created forums for leaders to explore topics ranging from enrollment and parent engagement to school operations, technology and AI. These programs have reached more than 1,000 Montessori leaders, creating a space to reimagine how technology can support the next generation of Montessori schools and educators.

The goal is not simply to promote technology. It is to build understanding, exchange ideas and give Montessori leaders a voice in shaping what gets built for their schools.

Building technology around Montessori practice

For Seldin, the value of the collaboration lies not in claiming that every challenge in Montessori operations has already been solved, but in the opportunity to keep building.

“What excites me is not a claim that every Montessori-specific technology problem has already been solved. It is how rapidly illumine is developing, the resources it can bring to the challenge, its experience building and supporting school-management technology at scale, and its willingness to listen to Montessori educators about what they actually need. That is one of the principal reasons for the collaboration.” said Tim Seldin, President of the Montessori Foundation.

The Foundation is bringing its experience and ideas to the development process, including perspectives on Montessori curriculum planning, classroom observation, parent communication, teacher support and narrative student reporting.

That exchange creates a continuous feedback loop: Montessori educators help identify the problems worth solving, technology creates new possibilities, and those possibilities return to the Montessori community for discussion, refinement and learning.

AI brings a new chapter

The collaboration is becoming even more consequential as artificial intelligence moves from a technology discussion into an operational reality. For Montessori schools, the question is increasingly not whether AI exists, but where it can create genuine value, where it should be used responsibly and where human judgment must remain central.

The opportunity is not to replace the teacher or the relationships at the heart of Montessori education. It is to reduce the administrative work surrounding them. A teacher can dictate an observation. AI can help organize it. Rough notes can become clearer reports. Information can be translated for families. Operational information can be turned into more useful reports.

The principle is simple: AI can draft. People decide.

The educator or administrator remains responsible for reviewing, correcting, approving and ultimately using the information. As AI develops rapidly, Montessori schools have an opportunity not merely to adopt new technology, but to help shape how it is applied within their unique educational context. illumine is committed to being part of that journey, building technology that advances what’s possible while preserving the human relationships and Montessori principles at the heart of education.

Montessori does not have to choose

The future of Montessori technology is not about replacing the principles that have made Montessori distinctive. It is about giving those principles a stronger operational foundation. The collaboration between The Montessori Foundation and illumine represents one approach to that future: bringing Montessori expertise and technology expertise together before building the solution.

The impact is already visible in schools moving away from fragmented systems, in a growing conversation around responsible AI and in technology being developed around Montessori workflows. Technology will continue to change rapidly. Montessori does not have to stand outside that conversation. It can help shape what comes next.

By bringing the people who understand Montessori together with the people building its technology, The Montessori Foundation and illumine are working toward a future where innovation does not move around Montessori, but is shaped with it.

About illumine
illumine is an AI-powered childcare management platform built for preschools, Montessori schools, and early education networks. It streamlines enrollment, tuition and billing, attendance tracking, regulatory compliance, and parent communication within a unified system powered by AI. Built to reduce administrative burden and improve operational efficiency, illumine supports childcare centers and multi-site programs across the United States and globally. For more information, visit www.illumine.app

View original content to download multimedia:https://www.prnewswire.com/news-releases/building-the-future-of-montessori-technology-how-the-montessori-foundation-and-illumine-are-shaping-what-comes-next-302903550.html

SOURCE illumine Technology Inc

Continue Reading

Technology

Omneky Harness Outperforms ChatGPT and Gemini on TasteBench Ad Score

Published

on

By

Omneky leads with an ad score of 8.8 compared with 7.6 for the next best tested configuration

SAN FRANCISCO, Oct. 13, 2026 /PRNewswire/ — Omneky, an AI advertising platform, today announced that its advertising harness earned the highest ad score in the TasteBench evaluation, scoring 8.8 and outperforming all six tested ChatGPT and Gemini configurations. The next-best configuration, ChatGPT using GPT-6 Astra at medium reasoning, scored 7.6.

Across the tested configurations, ChatGPT ad scores ranged from 7.3 to 7.6 and Gemini scores ranged from 6.8 to 7.2. Omneky led the nearest competitor by 1.2 points. The comparison covered 52 ad requests across four brands, producing 364 ads for evaluation.

The results highlight the role of the harness in creative production. A harness is the software surrounding a model that supplies context and directs its work. Omneky’s advertising workflow reads brand information, develops art direction, provides product and logo references, and selects an image model. TasteBench compares the ads produced by these complete workflows.

“Creative is one of the most important levers for growth, and the quality of the finished ad matters to every marketer,” said Hikari Senju, founder and CEO of Omneky. “We built Omneky’s harness to bring brand knowledge and creative direction into every generation. TasteBench gives marketers a way to compare the quality of what these systems actually produce.”

Omneky also led the separate readiness measure: 60% of its ads passed every applicable check, compared with 13% for the next-best configuration. This supporting metric measures whether an ad meets all of the benchmark’s required checks.

Omneky conducted the evaluation with three anonymous judges, applying the same standard to each tested system. ChatGPT and Gemini runs used APIs with web search and image generation; app spot-checks produced similar results. The findings assess generated creative and do not measure campaign returns.

Explore the results and side-by-side examples at www.omneky.com/tastebench.

About Omneky
Omneky is an AI advertising platform that helps businesses create image and video ads, review campaign performance and launch campaigns. Its agent brings these workflows into a single conversation. Learn more at www.omneky.com.

Media Contact: Hikari Senju | 424652@email4pr.com | (415) 236-2333

View original content to download multimedia:https://www.prnewswire.com/news-releases/omneky-harness-outperforms-chatgpt-and-gemini-on-tastebench-ad-score-302907011.html

SOURCE Omneky Inc

Continue Reading

Trending