Scaleway Generative APIs: set up a Paris-hosted model in 10 minutes
Scaleway Generative APIs serve open models from Paris through an OpenAI-compatible API. Here is how to get the key, test it and connect it to Locul.
Quick answer
Create a key in the Scaleway console under IAM & API keys > API keys > Generate API key, and copy the secret key, which is shown once. The base URL is https://api.scaleway.ai/v1. Test it with a curl call to /chat/completions and the model glm-5.3. In Locul, open Settings > Model > Open models, pick Scaleway, paste the key and press Test. Scaleway runs these models in Paris and applies zero data retention by default.
Who is Scaleway Generative APIs a good fit for?
Scaleway suits firms that want open models run by a French cloud in Paris, with prompts not kept and not used for training.
Scaleway is a French cloud provider, part of the Iliad group. Its Generative APIs are a serverless service: you pay per use, there is no server to manage, and the API speaks the same format as OpenAI's, so Locul talks to it directly. Generative APIs run in one region today, Paris (fr-par). Scaleway has other regions such as Amsterdam and Warsaw for its other products, but not for Generative APIs.
"By default we apply a Zero Data Retention Policy"
Scaleway docs, Generative APIs data privacy
The same page says Scaleway does not read or reuse your prompts and answers, and that they are not used to train models. The one exception: if a request harms the service, Scaleway may keep data for up to two weeks to investigate and fix the issue.
"Your data is not used for training, retraining, or improving the base models."
Scaleway docs, Generative APIs data privacy
| Provider | Country | Where prompts run | Prompts kept? | Used for training? | Pick in Locul |
|---|---|---|---|---|---|
| Scaleway (this guide) | France | Paris (fr-par) | Not kept by default | No | GLM-5.3 |
| Mistral AI | France | EU endpoint, api.eu.mistral.ai | Zero retention on request | Free plan may; opt out or use a paid plan | Mistral Medium 3.5 |
| OVHcloud | France | Gravelines | Not stored | Not stated; data not stored or shared | Qwen3.5 397B |
| IONOS | Germany | Berlin | Never logged | No | Qwen3.5 397B |
| Infomaniak | Switzerland | Swiss data centres | Not recorded | No | Qwen3.5 397B |
| Opper | Sweden | Stockholm gateway, EU model routes | Usage metadata only | No | DeepSeek V4 Pro (TensorX, EU) |
How do I create a Scaleway API key?
Open the Scaleway console, go to IAM & API keys, generate a key and copy the secret key before you close the window.
Sign in to the console
Go to console.scaleway.com. A new account needs a payment method and a verified identity before Generative APIs answer.
Open IAM & API keys
Click your organisation name in the top-right drop-down and choose IAM & API keys, then the API keys tab.
Generate the key
Click + Generate API key. Choose the bearer (yourself, or an IAM application if you want a key that is not tied to one person), add a description such as "Locul, Camille's Mac", and pick an expiry. When asked about Object Storage, you can skip it. Click Generate API key.
Copy the secret key
Scaleway shows an Access Key ID (starts with
SCW) and a Secret Key. Locul needs the Secret Key. Scaleway calls this screen your only chance to save it, so paste it straight into Locul or your password manager.
Try a model first. In the console, the AI section > Generative APIs lists every serverless model with a Try model playground, handy for a first look before you connect Locul.
What is the Scaleway Generative APIs base URL?
The base URL is https://api.scaleway.ai/v1, and the chat endpoint is /chat/completions under it.
Run this from a terminal to check the key before you open Locul. Replace the placeholder with your secret key.
# paste your key into the variable first, never into a shared file
$ export SCW_SECRET_KEY="paste-your-key-here"
$ curl https://api.scaleway.ai/v1/chat/completions \
-H "Authorization: Bearer $SCW_SECRET_KEY" \
-H 'Content-Type: application/json' \
-d '{"model":"glm-5.3","messages":[{"role":"user","content":"Reply with one word: ready"}]}'A working key returns JSON with a choices list:
{"object":"chat.completion","model":"glm-5.3",
"choices":[{"index":0,"message":{"role":"assistant","content":"Ready"},"finish_reason":"stop"}],
"usage":{...}}A 401 means the key is wrong, and a 404 on the model usually means a typo in the model name. The errors table below covers both.
Which Scaleway model should I pick in Locul?
Pick GLM-5.3: it has a 256k-token context window and handles the tool calls Locul's bots use to read files and memory.
| Model ID | Context | Use it for |
|---|---|---|
glm-5.3 | 256k tokens | Recommended. Long contracts, many files at once, bots with tools |
qwen3.5-397b-a17b | 250k tokens | A strong second choice; try it if GLM-5.3 phrasing does not suit you |
mistral-medium-3.5-128b | 180k tokens | A French-made model, in preview at Scaleway |
glm-5.2 | Retiring on 16 November 2026, then routed to GLM-5.3. Do not start on it |
Scaleway also hosts gpt-oss-120b, but tool calls for it only work through its Responses API, not chat completions, so Locul keeps it out of the main list.
How do I add Scaleway to Locul?
Open Settings > Model > Open models, pick Scaleway under In your region, paste the secret key and press Test.
Open Settings, then Model
In the Locul desktop app, open Settings and choose Model.
Choose Open models
Pick Open models (marked Recommended), then find Scaleway under In your country. Each row shows where the provider runs and whether it trains on your data.
Pick the model
Under Model, choose GLM-5.3. Locul tags it Recommended.
Paste the key
Under Access key, paste the key into the field and press Save and test. The Get a key link opens the right page at Scaleway if you have not made one yet.
Run the connection check
Under Connection check, press Test. Ready means chat and tools both work. Ready with limits means the model can chat but could not use tools in the check, so pick another model for bots that read files.
Where the key lives. Locul saves it in your Mac's Keychain. It never shows the key again and sends it nowhere except Scaleway.
Why does the Scaleway connection check fail?
Almost always because the Access Key ID was pasted instead of the Secret Key, or the account has no payment method yet.
| Code | Locul says | What to do |
|---|---|---|
| 401 | "Scaleway did not accept your access key." | Paste the Secret Key, not the Access Key ID that starts with SCW. If the secret is lost, generate a new key; Scaleway cannot show it again. |
| 404 | "Scaleway does not offer glm-5.3 for your account." | Check the model is still listed on Scaleway's supported models page. Retired models such as glm-5.2 after 16 November 2026 stop working. |
| 402 | "Scaleway says your account has no credit left." | Add or fix the payment method in the console under Billing, and finish identity checks if Scaleway asks for them. |
| 429 | "Scaleway is asking us to slow down." | Generative APIs have per-model rate limits. Wait a minute and press Test again, or ask Scaleway support to raise the limit for your organisation. |
| No answer | "Could not reach Scaleway." | Your network blocks api.scaleway.ai. Ask IT to allow HTTPS to that address. |
| Format | "That does not look like a whole access key." | The key is one long line with no spaces. Copy it again in one go, without a line break at the end. |
What does a real request on Scaleway look like?
A messy, everyday job with client names in it, the kind of work this setup is for.
Here is the sort of message an associate at a Lyon law firm might send to a Locul bot running on Scaleway:
The lease, the amendment and the landlord's letter are processed in Paris and not kept afterwards. The bot answers in English and leaves the French clauses as they are.
How fast and how good is Scaleway in Locul?
We are timing real office tasks on Scaleway with the recommended model, and the results will be published here.
Results coming
Response times, tool use and answer quality on real office tasks, measured on Scaleway. Coming soon.
What other in-country providers work with Locul?
Five more EU and Swiss providers, each with its own guide.
Mistral AI
The French model maker's own API for Mistral models.
Mistral AI guide →FROVHcloud
AI Endpoints serving open models on French cloud.
OVHcloud guide →DEIONOS
AI Model Hub running in German data centres.
IONOS guide →CHInfomaniak
AI Tools hosted in Swiss data centres.
Infomaniak guide →SEOpper
Swedish platform that routes to models hosted in the EU.
Opper guide →Frequently asked questions
Is Scaleway Generative APIs free?
Scaleway gives new accounts some free usage to start, then charges per million tokens. Rates change, so check Scaleway's pricing page for current figures.
What is a generative API?
It is a web address you send text to and get generated text back from, without running the model yourself. Scaleway Generative APIs host open models such as GLM-5.3 and Qwen3.5 and charge for what you use.
Does Scaleway keep my prompts?
No, not by default. Scaleway applies a zero data retention policy to Generative APIs. It may keep data for up to two weeks only if a request harms the service and needs investigating.
Can I use Scaleway Generative APIs in Amsterdam?
Not today. Generative APIs run only in the Paris region, fr-par. Scaleway's Amsterdam and Warsaw regions host other products.
Is the Scaleway API compatible with OpenAI?
Yes. It follows the OpenAI chat completions format at https://api.scaleway.ai/v1, so any OpenAI-compatible app, Locul included, can use it with a Scaleway key.
Sources
Every command and quote on this page comes from these primary sources, checked on 24 October 2026.