Choose OVHcloud AI Endpoints for the lowest list prices on small open models inside a French cloud you may already buy. Choose Munito AI for the newer large models, an Anthropic-compatible API, OCR of PDFs by the page, decision models, and a deployment in your own cloud or on-premises.
Prices and models checked on 2026-11-07. Prices change, so check them at the source.
At a glance
| Munito AI | OVHcloud AI Endpoints | |
|---|---|---|
| Where it runs | EU, US or global | EU |
| Prompts kept | No | No |
| API formats | OpenAI and Anthropic | OpenAI |
| OCR | PDFs and images, per page | Through a vision model |
| Decision models | Yes | No |
| Your own cloud | Yes | No |
| On-premises | Yes | No |
| Catalog | 8 chat models and OCR | Chat, vision, embeddings, speech |
Prices in the EU
Standard tier, per million tokens, input / output. Munito AI in USD, OVHcloud in EUR.
| Model | Munito AI | OVHcloud |
|---|---|---|
| Qwen3.8-27B | $0.20 / $2.70 | €0.40 / €2.70 |
| DeepSeek-V4-Flash | $0.11 / $0.28 | Not offered |
| DeepSeek-V4.1-Flash-1017 | $0.30 / $1.20 | Not offered |
| GLM-5.3-Flash | $0.15 / $0.50 | Not offered |
| Gemma 4 12B | $0.11 / $0.33 | Not offered |
OVHcloud also serves gpt-oss-120b at €0.08 / €0.40 and gpt-oss-20b at €0.04 / €0.15, which Munito AI does not list publicly, and Qwen3.5-397B-A17B at €0.60 / €3.60, which is larger than anything in our public catalog. Qwen3.8-27B is the one model both serve: our input price is half of theirs, and the output price matches.
The catalog is the real difference, not the rate card. OVHcloud leads on breadth of small models and on guard and embedding models. Munito AI leads on the newest large open models, where a reasoning or a long-context workload lives.
How to switch
Both speak the OpenAI API. Change two lines:
from openai import OpenAI
client = OpenAI(
base_url="https://api.eu.munito.ai/inference/v1",
api_key="munitoai_...",
)
The same key also answers the Anthropic Messages API, so an Anthropic-shaped agent needs no proxy.
Where OVHcloud is stronger
- Lower list prices for small models, including gpt-oss-20b at €0.04 input.
- A wider catalog by kind: embeddings, Whisper speech, guard models and image generation beside chat.
- One French cloud for compute, storage and inference, on one invoice, with sovereignty messaging a public buyer recognises.
- Free models for guard and image work, which keeps a prototype at zero.
Where Munito AI is stronger
- The newest large open models: DeepSeek-V4-Flash, DeepSeek-V4.1-Flash-1017 and GLM-5.3-Flash, each with a 1M-token context.
- The Anthropic Messages API beside the OpenAI Chat Completions and Responses APIs.
- EU, US or global processing on one key, with the region named in the
munito-regionheader. See Regions. - A deployment of your own: managed, inside your cloud account, or on-premises.
- OCR billed per page at $0.0044, so a document budget is a page count and not a token estimate.
- Decision models: pplx-decider-v1.2 and Cygnet answer yes or no, choice and score questions through the Decisions API.
- A cached-input price as low as $0.0060 per million tokens on DeepSeek-V4.1-Flash-1017, which a long system prompt turns into real money.
What a security review will ask
Both run in the EU, so the first question is answered either way. Ask for the rest in writing: the DPA, the sub-processor list, the retention period, and what happens to a prompt on an error path. Ours are at legal.munito.ai/dpa and trust.munito.ai/subprocessors. Our SOC 2 audit and ISO 27001 certification work are in progress; OVHcloud holds ISO 27001 and HDS today, which decides the matter if your checklist needs a certificate this quarter.
Sources
OVHcloud, checked 2026-11-07: AI Endpoints catalog.
Munito AI: Models and prices, Regions, Data privacy, Security.