NemoRouter vs Portkey vs LiteLLM vs Helicone — the 2026 feature-gating audit

The wedge claim: NemoRouter is the only LLM gateway that gives every customer all enterprise features — guardrails, A/B tests, prompt management, evals, budgets — free for life, with every major LLM provider behind one API key. Tiers vary the platform fee (4% / 2% / 0%); they never lock features.

There are two ways an LLM gateway can charge you. The first is feature gating — the routing is free or cheap, but guardrails, evals, A/B tests, prompt management, and per-team budgets sit one tier up, then another tier up, then "contact sales". The second is platform-fee gating — every feature is on every tier; what varies is the percentage on top of provider cost.

Four of the five gateways teams actually shortlist in 2026 do the first thing. NemoRouter does the second. This post is the audit: we walk the public pricing pages of Portkey, LiteLLM, Helicone, and OpenRouter, and we mark — feature by feature — which tier you have to be on to use that feature today. Every claim cites the vendor's own public docs or pricing page with a 2026-05-16 audit timestamp.

We are explicit about where NemoRouter sits — that is the wedge: every governance feature you would otherwise pay an enterprise contract for is included on Tier 1 (PAYG, 4% platform fee) the moment you sign up.

Five-minute path: the gating matrix, what each vendor actually charges for, the math for a $5k/mo team.

What "feature gating" means in this category

A modern LLM gateway has three layers:

1. Routing layer — model alias, fallback, weighted A/B, OpenAI-shaped /chat/completions, Anthropic-shaped /messages. Commodity. Every gateway ships this.
2. Governance layer — guardrails (PII / jailbreak / regex), per-team budgets + virtual keys, prompt management with versioning, eval pipelines, RBAC + multi-tenant isolation.
3. Provider-economics layer — pooling customer traffic into provider reservations (Azure OpenAI PTU, Vertex GSU, Bedrock Provisioned Throughput) so the gateway's margin doesn't have to come from the platform fee on the way in.

Every gateway gives you Layer 1 cheaply. The question is what each gateway does with Layer 2 — bundle it, or sell it as the upgrade path. NemoRouter bundles all of Layer 2 on Tier 1 and earns Layer 3 economics on Tier 3 ($1,200/yr prepay, 0% platform fee) as customers scale. Competitors monetize Layer 2 by tier. The matrix below is just that decision, drawn out.

The gating matrix

Read this as: "on which tier of this vendor does this feature first appear without a sales call?" Where a vendor distributes a feature across multiple tiers, we cite the lowest tier on which it appears at all. Every cell is sourced from the vendor's own public pricing or documentation page; Sources section at the bottom carries the URLs and audit dates.

OpenRouter, Portkey, LiteLLM, and Helicone are trademarks of their respective owners. NemoRouter is not affiliated with or endorsed by any of these vendors. Every row is sourced from each vendor's public pricing or documentation page on 2026-05-16; if any has changed, email hello@nemorouter.ai and we'll update.

Five rows of the matrix are governance features. On NemoRouter, all five are on Tier 1. On Portkey, Helicone, and LiteLLM Cloud, at least three of the five sit on a paid plan or a sales call. On OpenRouter, four of the five are not documented as product features at all — OpenRouter is, by its own framing, a routing product.

What each vendor actually charges for

It is fair to ask: what is each gateway monetizing? The answer differs across the four — and that difference is what makes the comparison interesting.

OpenRouter — a 5% platform fee on credits

OpenRouter charges a 5% platform fee on credit purchases on top of provider pass-through. The product surface is routing: model catalog, fallback chains, an OpenAI-shaped endpoint. Guardrails, evals, A/B testing, and per-team budgets are not documented as product features. If you need any of those, the answer in 2026 is to layer a second tool on top of OpenRouter — typically Portkey or Helicone — and pay for that separately.

Verdict: best fit for hobbyists and prototypes where governance is genuinely DEFER. Anything else, you are paying 5% to do routing and then paying again for the layer that controls cost and quality. The deeper head-to-head is in the OpenRouter alternative comparison.

Portkey — tiered subscription with governance on Pro and Enterprise

Portkey runs a tiered subscription. On the free tier you get routing and some observability; guardrails, prompt management, evals, A/B testing, and per-team budgets are distributed across the Pro and Enterprise plans. Annual prepay is an Enterprise contract that goes through sales.

Verdict: strong feature set, but governance is the monetization lever. If you need three or four of the Layer-2 features, you are buying Pro or Enterprise. That is a fine business model — it is just not the same model as ours, and a $5k/mo team should price both before committing.

LiteLLM — OSS for free, Cloud and Enterprise for the managed path

LiteLLM is genuinely open source. If you self-host the OSS proxy, you get routing, guardrails configuration, virtual keys, and observability for $0 — modulo the operational cost of running it. LiteLLM Cloud and Enterprise are the managed-service path, and that is where the gating lives: governance and SSO and team budgets at scale sit on the paid contract.

Verdict: the right answer for teams who want to self-host. It is the only credible OSS option in this comparison. The trade is operational: you run the gateway, you handle the upgrades, you write the alerting. For mid-market SaaS adding AI as a side feature, that operational tax usually outweighs the saved subscription. For platform/infra engineers who already run Kubernetes and want air-gap, it is the right call. See the buyer's guide § Buyer C for the full decision tree.

Helicone — observability-first, with governance on Pro and Enterprise

Helicone is the observability product in this matrix. It does routing, but the primary surface is per-request logs, traces, attribution, cost rollups, and dashboards. Guardrails and eval pipelines sit on the Pro and Enterprise tiers; per-team budgets and virtual keys at scale are an Enterprise concern.

Verdict: best fit when observability is your #1 problem and routing is #2. If guardrails, evals, and budgets all matter at the same time, you are buying the higher tier — at which point compare the annual cost line-for-line with Tier 3 below.

NemoRouter — platform-fee gating, not feature gating

NemoRouter charges a platform fee — 4% on Tier 1 PAYG, 2% on Tier 2 ($100/mo minimum), 0% on Tier 3 ($1,200/yr prepay). All Layer-2 features (guardrails, A/B tests, prompt management, evals, per-team budgets, RLS multi-tenancy) ship on Tier 1. The tier you choose changes the percentage we take on top of provider cost; it does not change the feature set.

The Layer-3 economic — pooling customer volume into Azure PTU / Vertex GSU / Bedrock Provisioned Throughput reservations — engages after we cross $10k ARR, and powers the Tier 3 0% platform fee as we grow. Customers continue paying retail PAYG provider rates; the reservation spread is our margin. Provider reservation pricing is documented up to 70% off PAYG on annual commits (linked in Sources).

Verdict: best fit for teams that want every governance feature today and a credible annual path to 0% platform fee. The wedge is in the first sentence of this post; if it stops being true, we have failed.

The math for a $5k/mo team

Pricing-page tables hide the punchline. So here is the same scenario priced four ways. Inputs: a team spending $5,000/month on provider tokens, who needs guardrails and per-team budgets today. We are not pricing seats, dashboards, or SSO add-ons — only the gateway. Each row uses the lowest tier on each vendor's published page that covers the two MUSTs (guardrails + per-team budgets); where the vendor distributes those features across plans, we use the lowest plan that has both.

We deliberately do not fill in live Portkey, Helicone, and LiteLLM Cloud numbers in the cells — those drift week to week and we will not invent them. Click the URLs in Sources, take 90 seconds, fill the row. The arithmetic is one cell each. If you would like us to run the exercise against your own invoices, that is what the 30-minute walk-through on /community is for; the full worked example is in the cost teardown.

Two observations that hold regardless of which numbers you write in:

• Tier 2 vs. Tier 3 inflects at ~$120k/yr of LLM spend. Below that, Tier 2 is cheaper if you do not need the higher RPM/TPM. Above it, Tier 3 wins on dollars and unlocks 1,000 RPM / 1M TPM. The cost teardown breakeven table draws this line.
• The "free tier" / "OSS" framings hide costs that show up later. OpenRouter free routing plus a second vendor for guardrails is usually pricier than Tier 1 once the second vendor's plan price clears the 1% platform-fee delta. LiteLLM OSS is $0 of subscription and a real ops bill — fine if you have a platform team, expensive if you do not.

When NOT to switch to NemoRouter

We would rather you skip the switch than switch on a math error. Three cases where another option is the right answer in 2026:

1. You need to self-host or air-gap. LiteLLM OSS is the call. We are a managed service today; custom deployments live under Enterprise.
2. Observability is your only problem and you do not need governance. Helicone is observability-first; it does that one thing very well.
3. You spend under $500/month and have zero compliance load. Any of the five works; pick the cleanest signup and move on. The fee delta is below noise at that scale.

Three cases where NemoRouter dominates regardless of which competitor we compare against:

1. **You need three or more Layer-2 features today and your monthly spend is

   $1,000.** Tier 1's 4% beats every paid plan we have seen at that spend — and Tier 3's 0% is uncontested at annual prepay.

2. You have multi-team or multi-customer attribution requirements. Per-team budgets + RLS are on Tier 1. No upsell, no Enterprise contract, no $50k floor.
3. You want a credible path to 0% platform fee within the budgeting horizon a CFO actually cares about. Tier 3 is $1,200/yr, today. Annual prepay funds annual provider reservations; the spread becomes our margin instead of yours.

The 5-minute audit you should run on your current gateway

If you are already on a gateway and reading this from the inside, run this before you switch anything:

1. Open your current gateway's pricing page. Find every feature you actually use. Note the tier each appears on.
2. Open /pricing on NemoRouter. Note that all of those features are on Tier 1.
3. Annualize your current spend (gateway subscription + per-seat + platform fee on the last 12 months of LLM spend). Write the number.
4. Compute NemoRouter Tier 3: $1,200 flat. Or Tier 2: $1,200 + 2% × annual spend. Or Tier 1: 4% × annual spend.
5. Pick the smallest number. That is your savings, in dollars, with zero feature loss. Add the migration cost (~half a day of engineer time per the two-line diff).

If the savings clear the migration cost, you have a deal worth closing. If they don't, stay where you are with a clean conscience — that is the audit working as intended.

Switching cost: the two-line diff

Migration into NemoRouter from an OpenAI-shaped gateway (which every competitor in this post exposes) is two lines for an existing OpenAI integration:

# Before — direct to OpenAI (or to any OpenAI-shaped gateway)
client = OpenAI(api_key=os.environ["OPENAI_API_KEY"])

# After — through NemoRouter
client = OpenAI(
    api_key=os.environ["NEMOROUTER_API_KEY"],
    base_url="https://api.nemorouter.ai/v1",
)

The Anthropic SDK accepts the same shape via its base_url parameter. Typical scoping: half a day for one engineer plus 24–72 hours of dual-run. Migration of prompt templates and guardrail configs is JSON in / JSON out — not a rewrite.

Try it on your own numbers

We auto-grant $5 in API credits on signup, no card required. Enough to route 5–10 production prompts across Tier 1, exercise the guardrails + per-team budgets UI on your real traffic, and decide whether the Tier 3 math holds before you sign anything.

→ Start free at nemorouter.ai/signup — Tier 1, $5 credit, no card. Mid-market SaaS or larger? Bring your last 90 days of LLM invoices to a 30-min walk-through (book through /community) and we'll redo the matrix against your actual spend.

See also

• OpenRouter alternative — every enterprise LLM-gateway feature, free for life
• How we cut LLM costs 60% without locking features behind an enterprise tier
• LLM gateway buyer's guide 2026
• From PAYG to Tier 3: when annual prepay beats per-call billing
• /pricing — canonical Tier 1 / 2 / 3 / Enterprise table

Sources

Vendor pricing/docs pages, verified 2026-05-16:

• OpenRouter pricing: openrouter.ai/pricing
• OpenRouter docs: openrouter.ai/docs
• Portkey pricing: portkey.ai/pricing
• Portkey docs: docs.portkey.ai
• LiteLLM docs: docs.litellm.ai
• LiteLLM GitHub: github.com/BerriAI/litellm
• Helicone pricing: helicone.ai/pricing
• Helicone docs: docs.helicone.ai
• Azure OpenAI Service pricing (PTU): azure.microsoft.com/en-us/pricing/details/cognitive-services/openai-service/
• Google Vertex AI generative AI pricing: cloud.google.com/vertex-ai/generative-ai/pricing
• AWS Bedrock pricing: aws.amazon.com/bedrock/pricing/
• Anthropic API getting started: docs.anthropic.com/en/api/getting-started