> Scaleway
Scaleway Generative APIs is a serverless, OpenAI-compatible inference service hosted only in Paris and billed per 1M tokens in euros. Published rates run from EUR 0.15/1M input and EUR 0.60/1M output for gpt-oss-120b, and EUR 0.15/EUR 0.35 for mistral-small-3.2-24b-instruct, up to EUR 1.80/EUR 5.50 for glm-5.2; whisper-large-v3 is metered per audio minute at EUR 0.003 with free output tokens, and embedding models are input-only at EUR 0.10/1M. Every new customer gets a free tier of 1,000,000 tokens, with billing starting at token 1,000,001.
// pros
- Free tier of 1,000,000 tokens for every new customer, stated directly on the product page: enjoy a free tier of 1,000,000 tokens and start paying only from the 1,000,001st token.
- Cached input is discounted on deepseek-v4-flash-0731 to EUR 0.08/1M against EUR 0.40/1M uncached, an 80% reduction stated in the table.
- Output charges are waived on embedding and speech models: qwen3-embedding-8b and bge-multilingual-gemma2 bill EUR 0.10/1M input with output Free, and whisper-large-v3 bills EUR 0.003 per audio minute with output Free.
- Zero Data Retention is the documented default: the privacy page states prompts, inputs and outputs are not collected, read or reused, only anonymised metadata and token counts, and customer data is not used to train the underlying models.
- Models are hosted on Scaleway's own infrastructure in Europe with no third-party service in the data path, and the product page cites first-token response times below 200ms for end users in Europe.
- OpenAI-compatible libraries and LangChain adapters work unchanged, and structured outputs plus native function calling are supported.
// cons
- The 1,000,000 free tokens are a one-off allowance rather than a monthly refresh, and their value depends entirely on which model consumes them.
- All Generative APIs are hosted in Paris only, and the FAQ states this may change in future, so non-European latency is not addressed on official pages.
- Official rate limits (tokens per minute, queries per minute) require a credit card on file and a passed KYC process; without both, the effective limits are lower.
- Prices are quoted in EUR, adding FX exposure to the token cost for accounts budgeted in USD.
- Model-to-model spread is wide: glm-5.2 at EUR 1.80/EUR 5.50 per 1M input and output is 12x the input rate of gpt-oss-120b or mistral-small-3.2-24b-instruct on the same table.
Use Scaleway when European data residency and zero-retention terms are hard requirements, and test first on the 1M free tokens; the cheapest rows (mistral-small-3.2-24b, gpt-oss-120b, pixtral-12b) make the most sense as defaults inside a multi-provider router.
written by Mike Fleming, Uprouter editorial · verified against official sources September 16, 2026
How we estimated this: the free allowance is converted to a dollar value using the provider’s own pay-as-you-go rates for the models a typical developer would reach for. When the allowance is metered in requests, tokens or minutes rather than dollars, we assume median usage of the cheapest capable model. Because assumptions are involved, the figure carries the est. flag — treat it as an order of magnitude, not a quote.
note: 1,000,000 free tokens per new customer, stated on the Generative APIs page; valued at the cheapest published input rate of EUR 0.15/1M (mistral-small-3.2-24b-instruct or gpt-oss-120b), which is roughly USD 0.16 - the grant is token-denominated, so actual dollar value varies by model.
| Plan | Type | Monthly | Input /1M | Output /1M | Note |
|---|---|---|---|---|---|
| Free tier (source-reported) | free | — | — | — | ~1,000,000 free tokens per new customer; billing starts from the 1,000,001st token; base rate limits require a valid payment method and increase automatically after identity verification; Batches API has no rate limit and a -50% discount. |
| 1M Free Tokens | free | — | — | — | One-time allowance of 1,000,000 tokens for all customers. |
| Generative APIs PAYG | payg | — | $0.220 | $0.650 | Estimated rates based on €0.2 input / €0.59 output per 1M tokens. |
Pricing normalized from public sources — always verify with the provider.
// model prices
| Model | Input /1M | Output /1M | Context |
|---|---|---|---|
| Mistral family | — | — | — |
| Qwen family | — | — | — |
| Pixtral 12B (2409) | — | — | — |
| Whisper Large v3 | — | — | — |
| Gemma 3 27B Instruct | — | — | 131k |
| Llama 3.3 70B Instruct | — | — | 131k |
$ Does Scaleway have a free tier?
Yes — Scaleway offers a free tier that we value at roughly $0.16 in free credits (estimate — derived from rate-limited usage at pay-as-you-go prices). Always confirm current limits on Scaleway's official pricing page — free tiers change without notice.
$ How much does Scaleway cost per 1M tokens?
Scaleway does not publish simple per-1M-token pricing (it may bill per request, per output, or via a subscription plan). Check their official pricing page for exact figures.
$ Is Scaleway safe to use?
Uprouter rates Scaleway as low risk. Low to medium, with the clearest policy position in this batch: the privacy page documents a Zero Data Retention default, no collection of prompt or output content, exclusion of customer data from model training, and hosting in Paris, France, where Scaleway states the service is not subject to extraterritorial laws such as the US Cloud Act. The retentions that do exist are bounded and disclosed: request content may be stored for up to two weeks only when traffic causes errors such as HTTP 500 or appears malicious, and anonymised usage data is kept up to six months. Pricing risk is structural rather than policy-based: a one-off token allowance, EUR billing and a single region. Risk ratings are editorial, evidence-based, and never influenced by affiliate relationships.
Raw community sentiment, shown unblended. Votes never feed into objective data fields like price, risk, or status — and never into sort order.
No published reviews yet — be the first.
// more_informationprice history · live status · risk & confidence · code snippets
price_history
No recorded changes yet.
live_status
current: upNo probes recorded yet.
Uprouter runs its own probes; best-effort, not a guarantee. Probe results depend on our network location and cadence — for production, always use the provider’s own status page.
risk_&_confidence
Low to medium, with the clearest policy position in this batch: the privacy page documents a Zero Data Retention default, no collection of prompt or output content, exclusion of customer data from model training, and hosting in Paris, France, where Scaleway states the service is not subject to extraterritorial laws such as the US Cloud Act. The retentions that do exist are bounded and disclosed: request content may be stored for up to two weeks only when traffic causes errors such as HTTP 500 or appears malicious, and anonymised usage data is kept up to six months. Pricing risk is structural rather than policy-based: a one-off token allowance, EUR billing and a single region.
Confidence reflects how much verifiable evidence (official docs, probe history, community reports) backs this entry. Risk is our editorial assessment of operator reliability and terms — not financial advice.
code_snippets
# Scaleway's own API — get a key from their dashboard
curl https://YOUR-PROVIDER-ENDPOINT/chat/completions \
-H "Authorization: Bearer THEIR_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "mistral-family",
"messages": [{ "role": "user", "content": "Hello" }]
}'Direct to the provider — Uprouter receives nothing; verify their current pricing before purchasing.
Ranked by neutral similarity signals only — live status, free-tier shape and free-credit proximity. Commercial relationships never influence this list.