Dedicated Inference
Reserved B200 and B300 GPU capacity that serves one open-weight model on an isolated, OpenAI-compatible endpoint with zero data retention and a 99.9% monthly availability SLA, billed by the GPU-hour and invoiced monthly upfront.
Find alternatives to Dedicated Inference
What it does
- Model inference
- GPU cloud
Sources
- Pricing
-
Page text fetched 2026-11-16 (curl, full text read); page quotes: "GPU time pricing."; "Monthly reservations are invoiced upfront.". GPU-hour rates are shown on the page with a monthly or yearly toggle; the digits render as animated counters, so no figure is quoted.
- Description
-
Official documentation · 16 Nov 2026
Page text fetched 2026-11-16 (curl, full text read); page quotes: "Reserved B200 and B300"; "Dedicated inference."; "Reserved capacity."; "Zero data retention"; "OpenAI compatible".
- What it does
-
Official documentation · 16 Nov 2026
Page text fetched 2026-11-16 (curl, full text read); page quotes: "Reserved B200 and B300"; "Dedicated inference."; "Reserved capacity."; "Zero data retention"; "OpenAI compatible". AI-related by Route A: the page sells reserved GPU capacity for serving open-weight models ("Production model serving").
- Sold within
-
Page text fetched 2026-11-16 (curl, full text read); page quotes: "morph dedicated create deepseek-v4-flash --balanced --yes"; "GPU time pricing.". A published GPU-hour plan list and a CLI purchase command (after the buyer approves price and terms) are on the page; no parent offering is named.
- Name
-
Official documentation · 16 Nov 2026
Page text fetched 2026-11-16 (curl, full text read); page quotes: "Reserved B200 and B300"; "Dedicated inference."; "Reserved capacity."; "Zero data retention"; "OpenAI compatible".
- URL
-
Official documentation · 16 Nov 2026
Page text fetched 2026-11-16 (curl, full text read); page quotes: "Reserved B200 and B300"; "Dedicated inference."; "Reserved capacity."; "Zero data retention"; "OpenAI compatible".
- Interfaces
-
Official documentation · 16 Nov 2026
Page text fetched 2026-11-16 (curl, full text read); page quote: "morph dedicated create deepseek-v4-flash --balanced --yes".
Also from Morph
-
Morph Fast Apply Model API
Specialised model API that merges AI-generated code edits into source files at 10,500+ tokens per second with 98% merge accuracy, in real-time and batched modes, with a self-hosted option.
-
WarpGrep API service
Code-search agent that searches a codebase in a separate context window, running grep, read and list-directory calls in parallel without embeddings, and returns only the relevant code to the coding agent that called it.
-
Compact API service
Context-compression API that shrinks a coding agent's chat history or code context by 50-70% at 33,000 tokens per second while keeping every surviving sentence verbatim, at $0.20 per million input tokens and $0.50 per million output tokens.
-
Reflex API service
Decision API that takes a state (text, JSON or up to 16 images) and typed choice, score and yes-or-no questions and returns a probability for every answer with zero output tokens, at $0.042 per million input tokens.
-
Model Router API service
Prompt classifier that scores a prompt's difficulty, ambiguity and domain in about 180 ms and tells the caller which model to send it to, at $0.005 per request.
Something wrong here? Send a correction — quote this product id: morph-dedicated-inference.