Comparison
Frontière AI vs Fireworks AI: the EU-sovereign alternative to US inference
Fireworks AI and Frontière AI both serve open-source inference, but they sit under different legal systems, and Fireworks AI offers capabilities we deliberately do not. Fireworks.ai, Inc. is a US company based in San Mateo, California, whose terms are governed by Delaware law; it provides fast serverless inference, fine-tuning and training. Its price list does not name Llama 3.3 70B, so we compare on GPT-OSS 120B, which both of us serve: its Standard tier publishes €0.13 ($0.15) per million input tokens before tax, against our €0.252 VAT included. Excluding VAT on both sides (Frontière AI: €0.21 in / €0.84 out per million), on this model Fireworks AI is cheaper on both input and output. What owning an advanced training stack does not change is who can compel access to it: a US company remains subject to US legal process under the CLOUD Act. Frontière AI is a French company: its 18 sovereign models run at EU providers with no non-EU capital control, the others are labeled fast access (US) before you call them, and it stores no prompts at all. Pick Fireworks AI for speed and fine-tuning; pick Frontière AI when you have to answer for jurisdiction rather than just for data location.
Side by side
Every row below comes from the other side's own published pages, checked on 2026-09-22 and linked at the end of this section — not from a third-party comparison.
| Criterion | Frontière AI | Fireworks AI |
|---|---|---|
| Operating company | FLEECE AI SASU | Fireworks.ai, Inc. |
| Headquarters | France | San Mateo, California, United States |
| Governing law | French / EU law | State of Delaware |
| US CLOUD Act exposure | Inference on the 18 sovereign models runs at EU companies; requests transit the gateway on Vercel (US company, Paris region, no prompt or response stored) | Yes — US company, regardless of where inference runs |
| How the catalog works | Curated catalog, 25 models live, each with a named primary provider | Direct provider with serverless inference (Standard, Priority and Fast tiers), fine-tuning and training |
| GPT-OSS 120B, per 1M tokens (in / out) | €0.252 / €1.008 incl. VAT (€0.21 / €0.84 excl. VAT) | €0.13 ($0.15) / €0.519 ($0.60) excl. tax |
| Prompt storage | Never stored by Frontière AI — only token counts and the amount charged. Sovereign providers: OVHcloud zero retention; Scaleway none except error or abuse requests, up to two weeks; fast-access models: the serving provider's own policy | Zero data retention by default on open models: prompts and generations are not logged to persistent storage without your opt-in |
| Jurisdiction visible per model | Yes — sovereign / fast access, on the card and in the API response | No EU data residency documented: the privacy policy names servers in the US; transfers covered by standard contractual clauses |
Prices published in USD, shown converted at 1 USD = 0.8655 EUR (rate of 2026-08-05) alongside the original figure. Fireworks AI's terms (section 9.1) say fees are stated exclusive of taxes such as VAT (checked on 2026-09-23). Frontière AI publishes in EUR, margin and VAT included — the amount actually debited — so the price verdicts on this page compare both sides excluding VAT (our price ÷ 1.2, French VAT).
About the price Fireworks AI's price list does not name Llama 3.3 70B, so this row compares GPT-OSS 120B, which both of us serve (Fireworks' Standard serverless tier). Fireworks also sells fine-tuning and training, which we deliberately do not; Frontière AI serves this model through Scaleway, an EU sovereign provider, inference only, with a fixed margin. You pay for a narrower scope with a clearer jurisdiction.
Sources, checked on 2026-09-22: docs.fireworks.ai — serverless pricing · docs.fireworks.ai — zero data retention · fireworks.ai — privacy policy · fireworks.ai — terms of service
Where Fireworks AI is the better choice
- Serverless inference in three tiers (Standard, Priority, Fast) lets you balance cost and latency per workload.
- Fine-tuning and training are first-class products; we only serve inference on a curated catalog.
- Zero data retention is the default on open models, documented in its data-handling guide.
- A broader catalog across model families and sizes.
Where Frontière AI is the better choice
- The operating company is French. There is no US parent, licensee or affiliate through which a US order could reach your data on the sovereign models.
- Every model carries its jurisdiction before you call it, in the catalog and in the API's own response — you never have to reconstruct who served a given request.
- API prompts and completions are never written to our database — only the token counts billing requires. The web console follows the same rule: conversations stay in your browser.
- Prices are published in euros, margin and VAT included, so the figure you read is the figure you are billed — US providers publish before tax.
How to choose, concretely
The two are not mutually exclusive, and for most teams the honest answer is to use both:
- Fine-tuning, custom weights, high-volume commodity inference: Fireworks AI does things we deliberately do not.
- Regulated, personal or client data — anything a DPO signs off on: pick the models whose operator jurisdiction is settled, not just the contractual safeguards.
- If you only remember one rule: data-centre location is a geography question, and a subpoena is a corporate-control question. Answer the second one first.
FAQ
Is Fireworks AI GDPR-compliant?
Fireworks AI documents zero data retention by default on open models, and standard contractual clauses for transfers out of the EEA. What none of that addresses is jurisdictional reach: Fireworks.ai, Inc. is a US company, so US legal process under the CLOUD Act applies to data it controls. Compliance paperwork and jurisdictional exposure are separate axes.
Is Frontière AI cheaper than Fireworks AI?
Fireworks AI does not list Llama 3.3 70B by name, so compare on GPT-OSS 120B: its Standard tier publishes €0.13 ($0.15) in / €0.519 ($0.60) out per million tokens before tax, against €0.252 / €1.008 VAT included on Frontière AI. Excluding VAT on both sides (Frontière AI: €0.21 in / €0.84 out per million), on this model Fireworks AI is cheaper on both input and output. Our prices include a fixed margin over what named providers charge us, and sovereign EU capacity is not the cheapest capacity on the market. The argument for Frontière AI is jurisdiction, not price.
Does Fireworks AI offer EU data residency?
Its privacy policy says it has servers for the service in the US and covers transfers from the EEA with standard contractual clauses; it does not document an EU-region option. If EU-region processing matters to your policy, that is worth confirming with them directly rather than inferring it.
Can I fine-tune models on Frontière AI?
No. Frontière AI scopes itself to inference on a curated catalog of open-source models. Fireworks AI, Together AI and others offer fine-tuning and training as first-class products. If fine-tuning is a core requirement, those providers will serve you better — and you can still use Frontière AI's sovereign models for the inference of regulated workloads alongside them.
See what we actually serve
25 models live, each with its publisher, licence, hosting jurisdiction and real price.