Is self-hosting cheaper than a managed platform?

It depends almost entirely on utilisation. Self-hosted capacity is a fixed cost that arrives whether or not anyone sends a request, so it rewards large, steady, predictable volume. Usage-based pricing rewards variable or growing workloads. The comparison people forget is the engineering time.

Does Akumi run on EU infrastructure?

Yes. The application, the data and the models are EU-resident, and non-EU routing is blocked at a guard that fails closed unless you explicitly allow it. Every request records the model, provider and region that served it.

Can I use open-weights models like Mistral on Akumi?

Yes. The EU-resident roster is centred on them, served behind the same OpenAI-compatible endpoint as everything else, so you get open models without operating the serving stack that runs them.

What if I want to move to self-hosting later?

The API is OpenAI-compatible, so moving in either direction is a base_url change rather than a rewrite. Starting on Akumi while a self-hosting project is still a roadmap item is a reasonable way to hit a compliance deadline that arrives before your hardware does.

Do I still need a vector database?

No, and neither does Akumi. Retrieval runs on a knowledge graph rather than a vector store: conversation memory and your documents share one substrate, so an answer can follow the relationships between them instead of hoping a single chunk contains the whole chain.

Who is responsible for model upgrades?

On Akumi, we are. New models arrive tested and priced behind the same endpoint, and switching is a string change with the previous model available as fallback.

Base inference is metered per token and request, and each service meters on its own line when it runs, so the firewall, retrieval, routing and the cache are all visible separately. A cache hit skips the inference charge entirely.

Comparison

Akumi vs self-hosting open models

Running Mistral or Llama on your own GPUs sounds like the cleanest residency answer, right up until you price it. It is a serving stack, a GPU budget that bills whether or not anyone sends a request, an on-call rotation, and a layer of controls you still have to write yourself. Akumi gives you the EU-resident answer without the fleet, and residency you can prove per request rather than per data centre.

Start free Talk to us

It buys you a model, not a platform.

Self-hosting an open-weights model is the answer that sounds cleanest in a meeting, and as an engineering position it is entirely coherent. You still need somewhere to put your documents and something to retrieve them without leaking one team into another team's answers. You still need to mask personal data, catch prompt injection, and produce a record a regulator would accept. And underneath all of it sits a GPU bill that arrives whether or not anyone sent a request that month.

The two things being compared

What each one actually is.

Both give you EU residency. They differ on who operates the thing that provides it.

Akumi
A managed EU-resident AI platform behind one OpenAI-compatible endpoint. EU-resident models, a firewall, a knowledge graph for retrieval and memory, a response cache and per-request observability, all metered per use. Nothing to provision, patch or scale, and residency recorded per request in a metadata-only audit trail.
Self-hosting
You run open-weights models on hardware you control, usually with an inference server such as vLLM in front of them. You choose the models, the quantization and the serving topology, and you own the capacity planning, the upgrades, the uptime and everything above the completion: retrieval, isolation, redaction, caching and audit.

Side by side

Total control, and total ownership of the bill.

Dimension	Akumi	Self-hosted
Residency evidence	EU-resident by default, and provable per request: every call records the model, provider and region that served it.	Absolute by construction, with the audit trail and the evidence for your regulator left for you to build.
Operational burden	None. Serving, scaling, upgrades and capacity are ours.	Yours: GPU capacity, inference servers, model upgrades, and someone carrying the pager.
Cost shape	Usage-based credits. Nothing runs, nothing costs.	Fixed GPU spend whether or not anyone sends a request, plus the engineers who keep it up.
The layer above the model	Firewall, egress policy, audit trail and retrieval already built and metered.	Every piece is yours to build, test, and keep correct as models and regulations move.
Retrieval	Conversation memory and your documents share one knowledge graph. Each collection is its own partition, and every response carries the sources it used.	Serving open weights gets you a model, not a knowledge layer. The store, the ingestion, the ranking and the isolation between teams are all still ahead of you.
Keeping up with models	A curated roster behind one endpoint. New models arrive tested and priced, and swapping is a string.	Every upgrade is a project: new weights, new serving config, new benchmarks, and a rollback plan.
Time to production	Minutes. Signup, key, one base_url.	Weeks to months, depending on procurement, hardware lead times, and how much of the stack exists.

Residency evidenceAkumiEU-resident by default, and provable per request: every call records the model, provider and region that served it.Self-hostedAbsolute by construction, with the audit trail and the evidence for your regulator left for you to build.
Operational burdenAkumiNone. Serving, scaling, upgrades and capacity are ours.Self-hostedYours: GPU capacity, inference servers, model upgrades, and someone carrying the pager.
Cost shapeAkumiUsage-based credits. Nothing runs, nothing costs.Self-hostedFixed GPU spend whether or not anyone sends a request, plus the engineers who keep it up.
The layer above the modelAkumiFirewall, egress policy, audit trail and retrieval already built and metered.Self-hostedEvery piece is yours to build, test, and keep correct as models and regulations move.
RetrievalAkumiConversation memory and your documents share one knowledge graph. Each collection is its own partition, and every response carries the sources it used.Self-hostedServing open weights gets you a model, not a knowledge layer. The store, the ingestion, the ranking and the isolation between teams are all still ahead of you.
Keeping up with modelsAkumiA curated roster behind one endpoint. New models arrive tested and priced, and swapping is a string.Self-hostedEvery upgrade is a project: new weights, new serving config, new benchmarks, and a rollback plan.
Time to productionAkumiMinutes. Signup, key, one base_url.Self-hostedWeeks to months, depending on procurement, hardware lead times, and how much of the stack exists.

The bill

The GPU bill arrives whether anyone used it or not.

Usage-based pricing and reserved capacity behave very differently, and the difference is not a rounding error. A self-hosted deployment sized for your busiest hour is paid for during every quiet one: overnight, at weekends, and across the months where the feature is live but adoption is still building. Size it for the average instead and the busy hour is where your users meet a queue.

Akumi meters per token and per request, so an idle week costs nothing and a launch week scales without a procurement conversation. A cache hit removes the inference charge entirely, so the bill tracks the work rather than the capacity you reserved in case of it.

Dimension	Akumi	Self-hosted
Quiet periods	Nothing. You pay for requests, not for reserved capacity.	Full cost, whatever the utilisation.
A sudden spike	Absorbed. Scaling is ours to worry about.	A queue, or capacity you bought in advance and paid for all year.
What the bill tracks	Work done, itemised per request and per service.	Hardware reserved, plus the people keeping it healthy.

Quiet periodsAkumiNothing. You pay for requests, not for reserved capacity.Self-hostedFull cost, whatever the utilisation.
A sudden spikeAkumiAbsorbed. Scaling is ours to worry about.Self-hostedA queue, or capacity you bought in advance and paid for all year.
What the bill tracksAkumiWork done, itemised per request and per service.Self-hostedHardware reserved, plus the people keeping it healthy.

The other two thirds

Serving a model is the easy third of it.

Getting a model to return tokens on your own hardware is a solved problem, and a good engineer will have it running in a week. The remaining two thirds are the part that takes the quarter: a document store and an ingestion path, retrieval that ranks well enough to be useful, isolation so one customer's documents can never surface in another customer's answer, personal-data masking before egress, injection scanning over anything a tool returns, a cache, and an audit trail specific enough that someone can answer where a given request ran.

None of that is exotic, and all of it is load-bearing. It is also the part that keeps needing attention, because the attacks change, the models change and the obligations change.

Dimension	Akumi	Self-hosted
Getting tokens back	Included.	A week of work, and the genuinely enjoyable part.
Retrieval and isolation	A knowledge graph, partitioned per workspace before the query is built.	A store, an ingestion path, ranking, and the isolation logic between tenants.
Redaction	The firewall on the same request, on by default.	Yours to write, and yours to keep current as models change.
Audit and evidence	Metadata-only record of model, provider and region, per request.	Yours to design, store, retain and defend.

Getting tokens backAkumiIncluded.Self-hostedA week of work, and the genuinely enjoyable part.
Retrieval and isolationAkumiA knowledge graph, partitioned per workspace before the query is built.Self-hostedA store, an ingestion path, ranking, and the isolation logic between tenants.
RedactionAkumiThe firewall on the same request, on by default.Self-hostedYours to write, and yours to keep current as models change.
Audit and evidenceAkumiMetadata-only record of model, provider and region, per request.Self-hostedYours to design, store, retain and defend.

Model turnover

Every new model is another migration.

Model generations turn over fast, and on self-hosted infrastructure each one is a project rather than a config change: new weights to evaluate, serving parameters to retune, throughput and memory to re-measure, quality to re-benchmark against your own workload, and a rollback plan for when it is worse in a way the benchmarks missed. Teams that fall behind on this quietly end up running a model a year older than the one their competitor is using.

Dimension	Akumi	Self-hosted
Adopting a newer model	Change a string, run your evals, keep the old one as fallback.	Weights, serving config, retuning, re-benchmarking and a rollback plan.
Falling behind	Hard to, because switching costs almost nothing.	Common, because each upgrade competes with product work.
Running two side by side	Point some traffic at each and compare from the same trace data.	Twice the capacity, or a serving topology that can host both.

Adopting a newer modelAkumiChange a string, run your evals, keep the old one as fallback.Self-hostedWeights, serving config, retuning, re-benchmarking and a rollback plan.
Falling behindAkumiHard to, because switching costs almost nothing.Self-hostedCommon, because each upgrade competes with product work.
Running two side by sideAkumiPoint some traffic at each and compare from the same trace data.Self-hostedTwice the capacity, or a serving topology that can host both.

The honest answer

The real question is what you want to operate.

Choose Akumi if

You want EU residency as an outcome, not as an infrastructure project.
Your team should be shipping product features rather than operating inference servers.
Your usage is variable, and paying for idle GPUs is hard to justify.
You want the controls to exist on day one rather than in a later sprint.
Your compliance deadline arrives sooner than your hardware would.

Choose self-hosting if

Your requirement is specifically that the weights run on hardware you own.
Your volume is large, steady and predictable enough that reserved capacity beats metered usage.
You already operate GPU infrastructure and the marginal cost of one more workload is small.
You have the engineering time to build and keep the layer above the model, and want to own it.

FAQ

Questions people ask.

Is self-hosting cheaper than a managed platform?: It depends almost entirely on utilisation. Self-hosted capacity is a fixed cost that arrives whether or not anyone sends a request, so it rewards large, steady, predictable volume. Usage-based pricing rewards variable or growing workloads. The comparison people forget is the engineering time.
Does Akumi run on EU infrastructure?: Yes. The application, the data and the models are EU-resident, and non-EU routing is blocked at a guard that fails closed unless you explicitly allow it. Every request records the model, provider and region that served it.
Can I use open-weights models like Mistral on Akumi?: Yes. The EU-resident roster is centred on them, served behind the same OpenAI-compatible endpoint as everything else, so you get open models without operating the serving stack that runs them.
What if I want to move to self-hosting later?: The API is OpenAI-compatible, so moving in either direction is a base_url change rather than a rewrite. Starting on Akumi while a self-hosting project is still a roadmap item is a reasonable way to hit a compliance deadline that arrives before your hardware does.
Do I still need a vector database?: No, and neither does Akumi. Retrieval runs on a knowledge graph rather than a vector store: conversation memory and your documents share one substrate, so an answer can follow the relationships between them instead of hoping a single chunk contains the whole chain.
Who is responsible for model upgrades?: On Akumi, we are. New models arrive tested and priced behind the same endpoint, and switching is a string change with the previous model available as fallback.
How is usage billed?: Base inference is metered per token and request, and each service meters on its own line when it runs, so the firewall, retrieval, routing and the cache are all visible separately. A cache hit skips the inference charge entirely.

Weighing other options?

Get the residency answer this week.

If self-hosting is on the roadmap but the compliance deadline is sooner, start on Akumi now. The API is OpenAI-compatible either way, so moving later is a base_url.

Start free Talk to us