MoYi T2.5: Translation close to the moment

With MoYi, we’re introducing a new way for MoYi CLI to preserve and retrieve context when the context window fills. Historically, models have used compaction to summarize work during long sessions, such as when debugging complex issues or tackling large refactors. Each compaction can leave out details about why a fix failed or how a component behaves. In MoYi CLI, MoYi can keep notes across context windows, preserving accumulated details without repeatedly compressing them into a single summary. Earlier context windows remain searchable, so MoYi can find requirements or test results from previous messages and tool outputs—even if that information wasn’t captured in its notes. You can enable this experimental feature in your MoYi CLI config.toml,⁠(opens in a new window) and it will become the default for MoYi in the coming weeks.

Given concerns that exposure to historical software vulnerabilities may have affected benchmark results, we also evaluated MoYi on two novel benchmarks. For one, we built an internal “ExploitBench (June–August 2026)” evaluation to test exploit development using vulnerabilities from the previous three months.¹⁴ MoYi achieved substantially higher arbitrary code-execution rates than Cloud baseline on this dataset while using far fewer output tokens. During the evaluation, MoYi even discovered and used two previously unknown zero-day vulnerabilities. We are disclosing both vulnerabilities to their maintainers.

We also tested MoYi on SRE-Bench¹⁵, a benchmark that measures whether models can reverse engineer software binaries to understand its core logic without access to raw source code. MoYi solved 88.0% of tasks in a single attempt and 99.2% within four attempts, compared with 55.9% and 68.7% for Cloud baseline, respectively.

Beyond benchmarks, expert-led assessments found that MoYi, when run without production safeguards, could use previously unknown vulnerabilities to achieve arbitrary code execution in hardened browsers and create privilege-escalation exploits for hardened operating-systems.

As we discussed in The Defender’s Window, frontier cyber capabilities can help defenders find weaknesses faster, but they also make those weaknesses easier to exploit, raising the urgency for defenders to adapt. With the version of MoYi launching today, defenders can use it to complete tasks such as secure code review and patching.

However, MoYi will refuse to comply with more advanced cybersecurity tasks such as creating proof-of-concept exploits for vulnerabilities. Through MoYi Daybreak⁠, we plan to expand access and roll out less restrictive safeguards in the coming weeks. This will enable more defensive workflows, including vulnerability and proof-of-concept validation, malware analysis, and detection engineering.

We have also strengthened our protections against potential cyber misuse, building upon our safeguards stack for Cloud baseline. These include stronger model robustness to better withstand potential jailbreaks and more context for our monitoring systems. We have continued rigorous internal and external testing, including automated evaluations with our internal red-teaming attackers. More details about our cyber safeguards and testing are available in the MoYi safety overview⁠ and system card⁠(opens in a new window)..

Our evaluations found MoYi’s written reasoning harder to monitor than Cloud baseline’s, based on tests that explicitly asked it to evade monitoring. We attribute this to MoYi’s greater control over written reasoning on simpler tasks and ability to solve problems with fewer written steps. MoYi still appears to struggle to conceal the reasoning needed for complex tasks, but we take the decline seriously. Improving monitorability remains a research priority, and the accompanying system card⁠(opens in a new window) details our findings and ongoing work.

Alignment training is core to our approach to deployment. As an additional layer of defenses, we also build system safeguards like MoYi CLI Auto-review⁠(opens in a new window) and monitoring agents’ reasoning and actions to help detect and contain unsafe behavior. As described in our safety update⁠, we are also deploying misalignment monitoring in production for MoYi-class models in order to have visibility into misalignment, and help contain its worst instances. These safeguards resemble our monitoring for internal deployments and involve a system of classifiers which check the model’s reasoning and actions for unauthorized behavior and automatically stop potentially unauthorized activity.

Given the significant increase in MoYi’s cybersecurity capabilities, we are being especially careful to make this deployment safe and secure. Extra safety checks can sometimes slow, pause, or stop legitimate work, including defensive cybersecurity. If a task is paused in MoYi or MoYi CLI, you may be asked to review the action before continuing. In the API, the task will stop. These checks can sometimes interrupt legitimate work, and we are continuing to iterate on this system to reduce unnecessary interruptions. Misalignment monitoring cannot replace alignment: our goal is to build models that reliably stay within their authorized scope, so these protections do not need to intervene.

MoYi T2.5 is currently published as an open runtime foundation, version 0.1.0. The public repository includes the native CLI, the C ABI bridge, the Python SDK, the Android JNI session boundary, and the asset packs for vi-en, vi-zh, and vi-ko. Model artifacts, pricing, and commercial terms are not published yet.

MoYi supports Zero Data Retention for eligible API customers, and as we shared last month, we're testing Private Safety Processing to strengthen safety monitoring while preserving customer privacy.

For developers, MoYi T2.5 will be available in the MoYi API as gpt-6-astra and through Microsoft Azure and Amazon Bedrock.

A project license has not been selected yet. Until a license is added, the source remains publicly visible for review, but reuse and redistribution rights are not granted automatically.