Rubytech Tests Russian LLM Cotype 3 on Chinese GPUs Matching NVIDIA H100 Performance
Rubytech has completed testing of the Russian large language model Cotype 3 from MWS AI on Chinese graphics processing units, demonstrating performance comparable to systems based on NVIDIA H100 accelerators.
The experiments were conducted on the Skala^r software-hardware complex designed for generative artificial intelligence workloads. In the test configuration, eight Chinese GPUs were deployed. When processing contexts of 27,000 tokens, the system achieved an average first-token generation time of approximately 8 seconds and inter-token latency ranging from 111 milliseconds.
According to Rubytech, these metrics are sufficient to support enterprise-grade AI services. Engineers performed extensive adaptations of drivers, the software stack, and orchestration mechanisms. In selected inference scenarios, these optimizations delivered speedups of 2 to 2.2 times relative to the default Chinese GPU environment.
The containerized architecture is expected to simplify model deployment and accelerate the transition from laboratory testing to production environments. The primary conclusion drawn from the trials is that corporate AI infrastructure can be constructed without exclusive dependence on NVIDIA accelerators.
This finding holds particular significance for government agencies, state-participated enterprises, and organizations operating critical information infrastructure that require localized model hosting, controlled infrastructure, and the ability to scale without supply-chain uncertainties associated with NVIDIA H100 GPUs.
Rubytech also indicated that Chinese GPUs can reduce total cost of ownership in specific use cases. The company stressed that the results represent an additional viable option rather than a full departure from NVIDIA technology. Future plans include testing newer generations of Chinese accelerators and expanding the range of supported Russian models and enterprise AI services.
Related articles
OpenAI Releases GPT-6.1 Terra: Cost-Effective Model Focused on Computer Control, Image Analysis and Practical Automation
OpenAI has launched GPT-6.1 Terra, a new model positioned between its high-end GPT-6 Astra and more affordable options in the GPT-6 family. The model emphasizes real-world computer interaction, visual understanding of documents and interfaces, and reduced error rates rather than raw reasoning benchmarks. It supports a 1.05 million token context window, image inputs, and tool use through the Responses API while cutting input token prices by half compared with its predecessor. Published benchmarks show substantial gains on Terminal-Bench 4.0, OSWorld 2.1 and Agents' Last Exam, with error rates dropping sharply at both low and maximum reasoning levels. OpenAI stresses that the model still requires human oversight for high-stakes actions and does not directly process audio or video. The release targets developers and enterprises seeking reliable automation of repetitive multi-step tasks at lower cost than flagship models.
Yandex Black October: How Architectural Flaws and Capacity Mismanagement Turned Data Center Losses Into a 11-Day Fintech Coma
In October 2026 three Yandex data centers in Central Russia were physically destroyed within days, triggering the collapse of Yandex Cloud and multiple core services. Between the first and third site failures engineers had 74 hours to restore quorum and rebalance workloads yet failed to do so because Control Plane components remained tied to the lost zone and spare capacity had been eliminated to cut costs. Core banking and BNPL systems entered Read-Only mode after losing quorum, forcing a complete stop-the-world reconciliation that lasted until 22 October. Backups stored inside the same regional S3 buckets proved unusable, resulting in a seven-day RPO and primary-key collisions after restoration. The incident exposed that Metro-Cluster designs had been incorrectly treated as full Disaster Recovery and that aggressive AP microservice patterns created logical split-brain conditions outside the database layer.
Astra Group Releases Low-Code Astra AI Platform for Building Corporate AI Agents in Closed Environments
Astra Group has launched the Astra AI Platform, a low-code constructor designed for creating corporate AI applications that operate entirely within closed circuits without reliance on foreign clouds. The platform targets small and medium businesses as well as large corporations, enabling rapid deployment of AI secretaries, technical support assistants, and other agents in days rather than months. It forms part of the broader Astra AI ecosystem and includes two main constructors: one for machine learning model training and deployment, and another for LLM-based language assistants, document search, and process automation. Users can assemble scenarios from visual blocks, debug them in real time, or extend functionality with custom Python code when needed. The solution supports RAG-based knowledge bases, integration with external tools such as CRM and ERP systems, and the coordination of multiple agents for distributed task handling. One demonstrated use case involves processing ten pages of engineering drawings and specifications in five minutes to prepare data for downstream systems. The product is already listed in the Russian domestic software registry and available to customers.
Multi-Tenancy Isolation in VictoriaMetrics, VictoriaLogs and Grafana: Building Client Separation in a Shared Stack
Voltir runs a single shared monitoring stack consisting of VictoriaMetrics, VictoriaLogs, Grafana and vmalert for all customers instead of deploying separate instances per tenant. Isolation relies on numeric account IDs enforced at every layer: nginx validates tokens against the platform API before proxying writes, Grafana data sources hard-code tenant accounts, and vmalert rules receive automatic account filters. The team discovered that a single malformed rule or template could break alerting for every tenant because vmalert loads the entire rules directory at once. They also found that Grafana automatically created personal organizations granting Org Admin rights, accumulating hundreds of orphaned organizations and users that consumed disk space and resources. A dedicated cleanup module plus hourly reconciliation now removes rules, organizations, series and users when a tenant is deleted, while isolation tests verify both data and API objects across tenants.