Comparison

FreeModel vs Portkey

Portkey assumes you already have provider keys and want to control what they do. FreeModel assumes you do not, and supplies the capacity itself. That is the fork.

The short answer

Pick Portkey if you bring your own provider accounts and need governance on top: budgets, RBAC, audit logs, guardrails, compliance certifications. It sits between you and providers you already pay.

Pick FreeModel if you do not have provider accounts and want an endpoint that works now, with free capacity first. You get less control and no enterprise governance — but you also do not need keys before you start.

Side by side

PortkeyFreeModel
Where model access comes fromYour own provider keys (BYOK only)Free capacity from upstream supply, paid fallback
Do you need provider accounts firstYes, for every provider you useNo
Core purposeGovernance and observability over spendSupply: an endpoint that works without setup
Free tier10K requests/month on the dev planFree-first routing by default
Paid plans$49/mo for 100K requests; $9 per additional 100KNone
Enterprise governanceRBAC, SSO/SCIM, audit logs, FinOps dashboardsNone
Compliance certificationsSOC 2, ISO 27001, GDPR, HIPAA; BAA signingNone published
Guardrails20+ deterministic and LLM-based checksNone
Deployment optionsManaged SaaS, hybrid (data plane in your VPC), self-hostHosted only
Anthropic-compatible endpointVia the universal APIYes, alongside OpenAI format

Portkey's published model coverage figures differ between their own pages, so this table states none. Check their docs directly if the count matters to you.

Where Portkey is the better choice

This is not a close call in several of these cases. If any of them apply, Portkey is the right pick and we are not.

1. You need governance, full stop

Budgets per team, RBAC, SSO with Okta, SCIM provisioning, audit logs, retention policies, FinOps dashboards. FreeModel has none of this and does not pretend to. If you are accountable for how a department spends on models, you need a layer that can answer "who spent what, and who approved it" — we cannot.

2. Compliance certifications are a hard requirement

SOC 2, ISO 27001, GDPR, HIPAA, BAA signing. If your procurement process requires these, the evaluation ends here. We publish nothing of the sort.

3. You need the data plane inside your own VPC

Portkey's hybrid deployment runs the gateway and data plane in your environment while they manage the control plane, so sensitive prompts never leave. FreeModel is hosted only. There is no configuration that changes this.

4. You want guardrails and observability built in

20+ guardrail checks, OpenTelemetry, 21+ metric dashboards. If you are running production traffic against models and need to see and constrain it, that toolkit exists in Portkey and does not in ours.

5. You already have provider accounts

If you are already paying OpenAI, Anthropic and Google directly, Portkey adds control without changing who you buy from. Switching to us would mean giving up direct billing for no governance gain.

Where FreeModel is the better choice

1. You do not have provider keys yet

This is the main one. Portkey is BYOK only: before it does anything for you, you need an account and a payment method at every provider you want to reach. FreeModel works without any of that.

2. Free capacity is the point, not governance

If what you actually need is "run this without paying yet", a governance layer does not help; it governs keys you do not have. FreeModel routes to free capacity first and falls through when it is gone.

3. You are one person, not a procurement process

Portkey's own positioning targets teams with compliance requirements. For a solo developer or a small project, the governance surface is overhead you will configure once and never look at again.

4. You want the agent layer, not the gateway layer

FreeModel sits under Aiglade, which ships four agent clients preset with task dispatch and an approval step. If you are orchestrating agents rather than governing API spend, that is the product you want.

Switching in either direction

From Portkey to FreeModel

base_url    https://freemodel.online/v1              (OpenAI format)
            https://freemodel.online/api/gateway     (Anthropic format)

You can stop supplying provider keys. That is the change.
Budget configs, guardrails and audit hooks have no equivalent here;
if you relied on them, migrate those decisions into your own code first.

From FreeModel to Portkey

1  Get provider keys (BYOK only, one per provider you use)
2  Add them in Portkey's dashboard or config
3  Point your client at Portkey's gateway URL
4  Recreate any budget or routing policy you had here as Portkey config

Note the billing change: after this, you pay providers directly.

What the supply side actually contains

The whole comparison rests on one question: who supplies the model access. Portkey supplies none; it routes to keys you add, so its coverage is whatever your provider accounts can reach. FreeModel supplies the list, and the list has a shape worth stating before you compare it against anything.

The public endpoint carries 199 ids. Of those, 77% are chat models, 9% generate images, and 23% cover everything else: video, speech synthesis, speech recognition, embeddings and rerank[1]. Six of the entries are routes rather than single models, and each carries a candidates count - how many models sit behind the name, in the order they are tried. That is the number that matters for an agent workload: depth behind a name, not depth across the list.

None of that is governance. It is inventory, and it is the half of the problem Portkey assumes you have already solved by opening accounts and payment methods at every provider you intend to use.

Deciding in three questions

The two products are not adjacent, so the evaluation is short. Ask these in order and stop at the first one that applies.

  1. Step 1: do you have provider accounts? If not, Portkey does nothing for you today: it is BYOK only, so every capability it has waits on a key and a payment method at a provider. FreeModel is the side that works before that step.
  2. Step 2: is compliance in your procurement path? SOC 2, ISO 27001, GDPR, HIPAA, BAA signing. If a reviewer requires any of these, the comparison ends here and Portkey is the answer. We publish nothing of the sort.
  3. Step 3: does the data plane have to stay inside your network? Portkey runs the gateway and data plane in your VPC in its hybrid deployment, with the control plane managed. FreeModel is hosted only, at every configuration.

Three noes and the answer is FreeModel; any yes and it is Portkey. There is no version of this where you need both for the same traffic, which is the useful thing to know before spending a week on a proof of concept.

FreeModel vs Portkey FAQ

Is FreeModel a drop-in replacement for Portkey?
No. The request-level change is small — point the client at https://freemodel.online/v1 or /api/gateway, swap the key, and but everything Portkey was doing for you has no equivalent here. Budgets, RBAC, guardrails, audit logs and retention policy would have to move into your own code before you switch, or be dropped deliberately.
What does Portkey do that FreeModel does not?
Per-team budgets and spend limits, RBAC, SSO with SCIM provisioning, audit logs, FinOps dashboards, 20+ guardrail checks, and certifications: SOC 2, ISO 27001, GDPR and HIPAA, with BAA signing. It also deploys as managed SaaS, hybrid, or self-hosted. We offer none of that, and a governance layer is not something you can approximate with a base URL change.
Do I still need provider keys if I use FreeModel?
No, and that is the structural difference. The platform key reaches the models without any provider account. Bringing your own keys is supported too: that provider then bills you directly and nothing is added on top. It is optional here, where with Portkey it is the only path.
Can FreeModel run inside my own VPC?
No. Requests pass through our gateway before they reach the provider, which is exactly what a data-residency requirement forbids. There is no self-hosted build and no hybrid option, so if the request path must stay inside your perimeter, no amount of configuration fixes it.
Which one should a single developer pick?
Usually FreeModel, for a boring reason: Portkey's governance surface is built to answer questions a team gets asked and a solo project does not. Configured once and never read again, it is overhead. The exception is a solo developer handling regulated data, where the certification requirement applies regardless of team size.

Last updated 2026-09-20. Portkey's pricing and features verified against their published docs on this date; their own pages report different model counts, so verify anything specific before relying on it.

Sources

Every third-party figure quoted on this site was read from the vendor documentation linked below, on the date stated. None of these links are affiliate links and none of these vendors paid to be listed. If a number here disagrees with the page it came from, the page is right — tell us and it gets corrected.

Where the percentages come from

Every share quoted on this site is counted from the live model list, not estimated. Divide the two counts in the last column, round to the nearest whole percent, and you get the number in the first.

ShareOf whatCount
77%are chat models153 of 199 listed entries
9%generate images18 of 199 listed entries
23%are not chat models46 of 199 listed entries
3%are modality routes6 of 199 listed entries
17%carry a context length34 of 199 listed entries

[1] Counted on 29 September 2026 from GET https://freemodel.online/v1/models, which needs no key and no account. Every row uses all 199 listed entries as the denominator, because the only thing the endpoint states for every entry is its modality. It no longer publishes per-model capability flags such as tool calling or reasoning, so this site quotes no shares for them. Fetch the list and recount whenever you want to check the arithmetic.