# TensorWave

**Canonical:** https://apis.io/providers/tensorwave/  
**Website:** https://tensorwave.com/  
**APIs profiled:** 5

TensorWave is a Las Vegas-headquartered AI cloud provider that builds and operates bare-metal GPU infrastructure exclusively on AMD Instinct accelerators (MI300X, MI325X, MI355X and MI455X) with AMD's open ROCm software stack. The company sells dedicated GPU nodes, high-speed network storage, and managed Slurm and Kubernetes clusters for AI model training, fine-tuning and inference, with observability, job scheduling, alerting and expert support layered on top. Access to the compute platform itself is delivered over SSH, Slurm and Kubernetes rather than a public REST control-plane API. TensorWave's public API surface is ScalarLM, the CC0-licensed unified training-and-inference stack it maintains and sponsors alongside RelationalAI, which exposes an OpenAI-compatible inference endpoint backed by vLLM together with a Megatron-LM training surface from a single deployment.

## Kin Score — 45.1 / 100 (developing)

Scored 2026-08-25 under rubric 0.14.0. Trend: flat (-0.7 from 45.8).

| Facet | Score |
|---|---|
| Discoverability | 74.1 |
| Contract Quality | 48.3 |
| Governance | 30.3 |
| Contract Governance | 30.3 |
| Operational Transparency | 28.9 |
| Developer Ergonomics | 54.2 |
| Commercial Clarity | 36.8 |
| Access Clarity | 36.8 |

## Agent readiness — 21.3 (agent-aware)

| Dimension | Value |
|---|---|
| Spec Presence | yes |
| Agentic Access | no |
| Reversibility Documented | documented |
| MCP Server | no |
| Auth Clarity | no |
| Idempotency | no |
| Error Semantics | verified |
| OpenAPI Examples | no |
| Rate Limit Signal | no |
| Event Surface Described | no |
| Agent Skills | derived |
| Well Known Catalog | no |
| Consent Identity | no |
| Agent Card | no |
| Dry Run Mode | no |
| Delegated Identity | no |
| Protected Resource Metadata | no |
| Dynamic Client Registration | no |
| Agentic Commerce | no |

## APIs (5)

- **TensorWave Generate API** — Queue-backed batch inference and worker coordination.
- **TensorWave Health API** — Health checks, service logs and metrics.
- **TensorWave Open AI API** — OpenAI-compatible inference endpoints proxied to vLLM.
- **TensorWave Slurm API** — Slurm scheduler status and job control.
- **TensorWave Training API** — Megatron-LM training jobs, chunked dataset upload, checkpoints and Hugging Face publishing.

## MCP servers (1)

- **TensorWave MCP Server**

## Security (4)

- **Tensorwave Authentication** — 0 schemes
- **Tensorwave Domain Security** — TLSv1.3 · HSTS · DMARC
- **Tensorwave Vulnerability Disclosure** — Hackerone · contact published
- **Tensorwave Trust Center** — SOC 2, ISO 27001, HIPAA

## Tags

Company, Artificial Intelligence, Machine-Learning, Cloud Computing, GPU, Infrastructure, Inference, Model Training, High Performance Computing, Data Centers

---

Profiled by [API Evangelist](https://apievangelist.com) and published on [APIs.io](https://apis.io/providers/tensorwave/). Scores are computed from the provider's own public artifacts under a published rubric.