LogoTensorFusion
  • Pricing
  • Docs
GPU Go ConsoleTensorFusion EE
LogoTensorFusion

Boundless Computing, Limitless Intelligence

GitHubGitHubDiscordYouTubeYouTubeLinkedInEmail
Product
  • Pricing
  • FAQ
Resources
  • Blog
  • Documentation
  • Ecosystem
  • Changelog
  • Roadmap
  • Affiliates
Company
  • About
Legal
  • Cookie Policy
  • Privacy Policy
  • Terms of Service
© 2026 NexusGPU PTE. LTD. All Rights Reserved.

Roadmap

Explore our future plans and upcoming features

Discuss on GitHub
Products
Task types
Backlog6

Mount multiple remote GPUs from different hosts

Aggregate remote GPUs across hosts and expose them as a single logical pool to workloads.

GPU GoTensor Netfeaturearchitecture

MetaX GPU support

Vendor integration for discovery, telemetry, and isolation modes where available.

Tensor Fusion EngineTensor OSecosystem

AWS Neuron support

Support Neuron devices for scheduling, monitoring, and isolation templates where applicable.

Tensor Fusion EngineTensor OSTensor Netecosystemarchitecture

TensorNet cross-cluster GPU scheduling

Schedule and route workloads across clusters/regions with "compute to data" policies and global quotas.

Tensor Netfeaturearchitecture

GPU Go cloud sync & multi-device

Cross-device GPU resource sync for GPU Go personal/team plans.

GPU Gofeature

AI model registry & preloading

Build your own private MaaS (Model-as-a-Service) with model caching and preloading.

Tensor OSTensor Netfeature
In progress5

Ascend NPU Soft-isolation

Production-ready limiter workflow and observability for Ascend NPU oversubscription scenarios.

Tensor Fusion EngineTensor OSfeatureecosystem
2026 Q1–Q2

AMD GPU Soft-isolation

Hook-based time-sharing isolation for AMD GPUs, aligned with TensorFusion quota + scheduler.

Tensor Fusion Enginefeatureecosystem
2026 Q1

Gang scheduling

First-class support for multi-vGPU / multi-accelerator workloads requiring atomic placement.

Tensor Fusion EngineTensor Netfeaturearchitecture

Topology-aware scheduling

Place workloads with awareness of NUMA/NVLink/PCIe/IB topology to maximize performance and stability.

Tensor Fusion EngineTensor Netarchitecture

Benchmark matrix

Standard benchmark suite across vendors, isolation modes, transport (Ethernet/RDMA), and frameworks.

Tensor Fusion Engineperformance
Released33

AMD GPU remoting

Remote GPU support for AMD GPUs with TensorFusion scheduling and telemetry.

Tensor Fusion EngineTensor OSfeatureecosystem
2026-01

Hygon DCU remoting

Remote GPU path for Hygon DCU devices with unified scheduling integration.

Tensor Fusion EngineTensor OSecosystemfeature
2025-12

NPU virtualization templates

Standardized partition/isolation templates for NPUs to accelerate onboarding and operations.

Tensor OSTensor Fusion Enginearchitecturefeature
2025-12

Heterogeneous device support

Support multiple GPU/NPU vendors in the same cluster with unified scheduling.

Tensor Fusion EngineTensor OSarchitectureecosystem
2025-12

Hard isolation: spatial-division sharing

Space-sharing mode for stronger isolation guarantees (no oversubscription).

Tensor Fusion Enginearchitecture
2025-11

Partitioned scheduling (MIG-like)

Hardware-partitioned isolation scheduling for MIG and similar technologies.

Tensor Fusion Enginearchitecturefeature
2025-11

Device controller

Dedicated controller for managing accelerator lifecycle and health.

Tensor Fusion Enginearchitecture
2025-11

Soft/hard/shared isolation modes

Three isolation modes for compute percent scheduling with different trade-offs.

Tensor Fusion Enginearchitecturefeature
2025-10

Elastic rate limiter

Adaptive compute throttling with PID controller for smooth resource sharing.

Tensor Fusion Engineperformancearchitecture
2025-10

VRAM hard-isolation

Strict memory enforcement for GPU workloads requiring hard memory limits.

Tensor Fusion Enginearchitecture
2025-10

GPU workload autoscaling

Auto-scale GPU workloads based on utilization and pending demand.

Tensor Fusion EngineTensor OSfeature
2025-09

Karpenter node expansion

Auto-expand GPU nodes when pods are pending, integrated with Karpenter.

Tensor Fusion EngineTensor OSfeatureecosystem
2025-09

GPU worker preemption

Preempt lower-priority GPU workers to improve scheduling fairness.

Tensor Fusion Enginefeature
2025-09

RDMA transport support

RDMA path for low-latency/high-throughput remote GPU access and scheduling.

Tensor Fusion EngineTensor Netperformancearchitecture
2025-08

Hypervisor health probes

Healthz/readyz APIs for hypervisor liveness and readiness monitoring.

Tensor Fusion Enginearchitecture
2025-08

Large-scale benchmark & optimization

Performance optimization for high GPU count clusters based on benchmarking.

Tensor Fusion Engineperformance
2025-08

GPUNodeClaim & Karpenter integration

Cloud vendor integration and Karpenter auto-scaling for GPU nodes.

Tensor Fusion EngineTensor OSfeatureecosystem
2025-07

Progressive migration from NVIDIA operator

Migrate from existing NVIDIA operator/device-plugin setups incrementally.

Tensor Fusion Enginefeature
2025-07

Kubernetes device plugin integration

Native K8s device plugin integration in hypervisor for standard resource management.

Tensor Fusion Enginearchitecture
2025-07

Hypervisor TUI monitoring

Real-time terminal UI for monitoring workers and GPU state.

Tensor Fusion Enginefeature
2025-07

Full-fledged NVIDIA remoting

Production-grade GPU-over-IP for NVIDIA, including Windows vGPU and Remote GPU.

Tensor Fusion EngineTensor OSfeaturearchitecture
2025-06

K8s scheduler framework refactor

Refactored to Kubernetes scheduler framework for advanced scheduling policies.

Tensor Fusion Enginearchitecture
2025-06

Alertmanager integration

Integrated alerting with Prometheus Alertmanager for GPU cluster monitoring.

Tensor Fusion EngineTensor OSecosystem
2025-06

Multi-GPU requests

Allow workloads to request multiple GPUs with model filtering.

Tensor Fusion Enginefeature
2025-05

Per-GPU UUID limits

Set CUDA limits per GPU using device UUIDs or indices.

Tensor Fusion Enginefeature
2025-05

Weighted scheduling

Weighted scheduler for fair GPU resource distribution.

Tensor Fusion Enginefeature
2025-05

Canary/gray rollout for TF Pods

Gradual rollout support for TensorFusion-enabled Pods.

Tensor Fusion Enginefeature
2025-04

CUDA memory hooks (cuMemCreate)

Hook CUDA memory APIs for strict memory limit enforcement.

Tensor Fusion Enginearchitecture
2025-04

TFLOPs-based resource limiting

Limit GPU resources based on TFLOPs for fine-grained control.

Tensor Fusion Enginefeature
2025-03

Distribution controls (maxSkew)

Control workload distribution across nodes with maxSkew parameter.

Tensor Fusion Enginefeature
2025-03

GPU temperature monitoring

Monitor GPU temperature for thermal management and alerting.

Tensor Fusion Enginefeature
2025-03

GPU metrics foundation

TFLOPs/VRAM metrics pipeline across controller and engine.

Tensor Fusion Enginearchitecture
2025-01

GPU pool management

Manage GPU resources as pools with component configuration.

Tensor Fusion Enginefeature
2025-01