Neysa Inference Endpoints
Dedicated single-tenant endpoints that serve open-source, open-weight or customer fine-tuned models on reserved NVIDIA or AMD GPUs, with OpenAI-compatible APIs, vLLM or SGLang engine selection per workload and hybrid pricing with scale-to-zero.
Find alternatives to Neysa Inference Endpoints
What it does
- Model inference
- Model hosting
Sources
- Pricing
-
Pricing page read 2026-11-16: the Included list names "Inference endpoints, at the GPU rate", and the page says "Every rate is published and billed to the minute, so the number you plan for is the number you pay." The endpoint page (https://neysa.ai/neysa-velocis/inference-endpoint/) adds "Hybrid pricing with scale-to-zero" and "Request a Demo" for a sized configuration; no price of its own is stated beyond the GPU rate, so only usage-based is recorded and no free model.
- Description
-
Official documentation · 16 Nov 2026
Page quotes: "Deployed on dedicated infrastructure, sized to your latency targets, and traffic; billed at predictable monthly spend."; "Run any open-source model with a HuggingFace checkpoint or bring your own fine-tuned weights, custom containers, or Helm charts."; "OpenAI-compatible endpoints"; "Per-workload kernel and engine selection (vLLM, SGLang, custom CUDA)"; "Hybrid pricing with scale-to-zero"; "NVIDIA configurations from L4 through Blackwell B300, plus AMD Instinct for teams that want a non-NVIDIA path."
- What it does
-
Official documentation · 16 Nov 2026
Page quotes: "Consistent, high-performance inference – more tokens per second, lower latency, and optimized throughput even under heavy workloads."; "Get dedicated single-tenant inference endpoints running on vLLM, deployed on reserved monthly GPUs for guaranteed availability and security and ability to customize every aspect of your endpoint"
- Sold within
-
The pricing page's "Included Add on" block lists "Inference endpoints, at the GPU rate" in the Included column beside Velocis's compute, VPC and ML tooling, a row that names this feature (standard 3: a platform-wide table proves nothing unless a row names the feature). Parent name copied from the sibling record Neysa Aegis LLM Shield.
- Name
-
Official documentation · 16 Nov 2026
Page heading: "Inference Endpoints: Real-Time Inference Tuned to Your Workload"; the vendor's page title is "Inference Endpoints - Deploy & Scale LLMs & AI Models". Written "Neysa Inference Endpoints" beside the sibling record Neysa Aegis LLM Shield, as the vendor-prefixed form; the page itself says only "Inference Endpoints". Own page, own heading, benchmark results and a pricing-page row.
- URL
-
Official documentation · 16 Nov 2026
Page text fetched 2026-11-16 with curl, full text read; the page is the Inference Endpoints entry of the Velocis menu, listed there as "Inference-as-a-Service" with its own page. Page title: "Inference Endpoints - Deploy & Scale LLMs & AI Models".
- Deployment
-
Official documentation · 16 Nov 2026
Page quote: "Deploy in your VPC, on-prem, air-gapped edge, or multi-cloud"; the endpoints also run on Neysa's own cloud ("Deployed on dedicated infrastructure"). VPC is read as private-cloud, on-prem as self-hosted, air-gapped edge as air-gapped.
- Interfaces
-
Official documentation · 16 Nov 2026
Page quote: "OpenAI-compatible endpoints"
Also from Neysa
-
Neysa Velocis Infrastructure service
Neysa Velocis is a GPU cloud to train, fine-tune and serve AI models, with networking, observability and governance for each workload and private model repositories inside the customer's VPC.
-
Neysa Aegis LLM Shield Platform
Sits between users and LLM endpoints to block prompt injection and jailbreaks, redact PII and enforce security policies on every request and response, with no change to application code.
Something wrong here? Send a correction — quote this product id: neysa-inference-endpoints.