Compare
ModelGate, Portkey, LiteLLM, Helicone and Cloudflare AI Gateway
All five sit between your app and your model providers. They differ in what they do with that position. This page compares ten criteria using each vendor’s own documentation, reviewed 7 October 2026.
Where ModelGate is different
- It works from your repository as well as your traffic: findings point at the line, and fixes open as pull requests you merge.
- For rule-like tasks it proposes plain code, tested against your model’s own recorded answers.
- Pricing is a base fee plus a share of savings, not logged requests or seats.
We found none of the first two in the reviewed documentation for the other four. That is not proof they don’t exist.
Where others are stronger
- Far more providers (100+ for LiteLLM and Helicone’s gateway, versus four).
- Semantic caching, load balancing and richer routing.
- Self-hosting, SSO, and compliance programs — ModelGate has no certifications.
If those matter most, one of them may fit you better.
How to read it. “Native + integrations” means built-in features plus third-party services you connect. “Not confirmed in reviewed documentation” means we found no mention — not that the feature is missing. ModelGate’s own “Limited” marks a real feature with a narrow scope, explained in the note. Planned ModelGate features are not counted.
| Criterion | ModelGate | Portkey | LiteLLM | Helicone | Cloudflare AI Gateway |
|---|---|---|---|---|---|
| Request and cost observability | AvailableEvery request logged, failed ones included, and priced against a versioned rate table. Four providers: OpenAI, Anthropic, Google Gemini, Azure OpenAI.Details: ModelGate documentation | AvailableLogs, traces, custom metadata and cost attribution across a universal API.Source: Portkey — Pricing (opens in a new tab) | AvailableLogs and spend per request, tracked by key, user, team and tag; 100+ LLMs.Source: LiteLLM docs — Proxy (AI Gateway) (opens in a new tab) · Source: LiteLLM — Enterprise (opens in a new tab) | AvailableRequest observability and cost tracking, with an HQL query language on paid plans.Source: Helicone — Pricing (opens in a new tab) | AvailableAnalytics on requests, tokens and cost, plus request logs.Source: Cloudflare AI Gateway docs (opens in a new tab) |
| Budgets and access controls | AvailablePer-project monthly spend cap (402), requests-per-minute per key (429), token ceilings (413); scoped API keys; agency roles and per-client caps. No SSO.Details: ModelGate security | Higher plans onlyRBAC and service-account keys on paid plans. Cost and token budget limits are listed for Enterprise and select Pro users.Source: Portkey — Pricing (opens in a new tab) · Source: Portkey docs — Budget limits (opens in a new tab) | AvailableVirtual keys with max budgets and RPM/TPM limits per key, user and team (needs a database). SSO, SCIM and RBAC listed under Enterprise.Source: LiteLLM docs — Budgets and rate limits (opens in a new tab) · Source: LiteLLM — Enterprise (opens in a new tab) | AvailableCustom rate limits, including cost-based quotas per user or segment (429 when exceeded).Source: Helicone docs — Custom rate limits (opens in a new tab) | AvailableRate limiting; budget-limit and rate-limit nodes in dynamic routing switch to a fallback when exceeded.Source: Cloudflare AI Gateway docs (opens in a new tab) · Source: Cloudflare AI Gateway — Dynamic routing (opens in a new tab) |
| Security guardrails | LimitedBuilt-in heuristic detectors: prompt-injection scoring, 17 secret signatures plus an entropy pass, 6 personal-data types. Detect on every plan; enforce on paid plans. No content-moderation categories.Details: ModelGate security | Native + integrations20+ built-in deterministic checks and LLM-based checks (including prompt injection), plus partner guardrails such as Aporia and Pillar.Source: Portkey docs — Guardrails (opens in a new tab) | Native + integrationsBuilt-in content filter and Presidio PII masking; many third-party guardrails (Lakera, Bedrock, Azure Content Safety…). Per-key guardrail control is Enterprise.Source: LiteLLM docs — Guardrails (opens in a new tab) | LimitedLLM Security (Meta Prompt Guard / Llama Guard) via request headers; the docs say it currently works with OpenAI models only.Source: Helicone docs — LLM Security (opens in a new tab) | AvailableGuardrails flag or block harmful prompts and responses (billed as Workers AI usage); DLP scanning.Source: Cloudflare AI Gateway — Guardrails (opens in a new tab) · Source: Cloudflare AI Gateway — Pricing (opens in a new tab) |
| Caching and model routing | LimitedExact cache only (temperature 0, no tools, 24 h). Routing moves small, simple calls to a fixed same-provider cheaper model; retry plus one secondary provider. No semantic cache or load balancing.Details: ModelGate documentation | AvailableSimple cache on all plans; semantic cache on select Enterprise plans. Load balancing, retries and fallbacks.Source: Portkey docs — Simple and semantic cache (opens in a new tab) · Source: Portkey — Pricing (opens in a new tab) | AvailableRedis, semantic (Redis, Valkey, Qdrant), S3/GCS, disk and in-memory caches; load balancing, routing and fallbacks.Source: LiteLLM docs — Caching (opens in a new tab) · Source: LiteLLM docs — Proxy (AI Gateway) (opens in a new tab) | AvailableExact-match cache (7-day default TTL); AI Gateway with routing and automatic fallbacks.Source: Helicone docs — Caching (opens in a new tab) · Source: Helicone docs — AI Gateway (opens in a new tab) | AvailableCaching; dynamic routing with conditional and percentage splits, retries and model fallbacks.Source: Cloudflare AI Gateway docs (opens in a new tab) · Source: Cloudflare AI Gateway — Dynamic routing (opens in a new tab) |
| Repository analysis | LimitedGitHub only. Reads the repositories you choose in memory and maps LLM calls to file and line (JavaScript/TypeScript and Python). Source is not stored.Details: ModelGate docs — GitHub connection | Not confirmed in reviewed documentationNo mention found in the pages reviewed. | Not confirmed in reviewed documentationNo mention found in the pages reviewed. | Not confirmed in reviewed documentationNo mention found in the pages reviewed. | Not confirmed in reviewed documentationNo mention found in the pages reviewed. |
| Integration and optimization pull requests | LimitedA setup PR that points OpenAI SDK clients at the gateway; one-line PRs for a model switch or a lower output limit; a PR that adds a verified code replacement. Always on a new branch.Details: ModelGate docs — GitHub connection | Not confirmed in reviewed documentationNo mention found in the pages reviewed. | Not confirmed in reviewed documentationNo mention found in the pages reviewed. | Not confirmed in reviewed documentationNo mention found in the pages reviewed. | Not confirmed in reviewed documentationNo mention found in the pages reviewed. |
| Deterministic code replacement | LimitedOnly for rule-like tasks (a few fixed labels, extraction, JSON reshaping, arithmetic, simple image checks) with 50+ recorded calls and prompt and response storage on.Details: ModelGate docs — code replacements | Not confirmed in reviewed documentationNo mention found in the pages reviewed. | Not confirmed in reviewed documentationNo mention found in the pages reviewed. | Not confirmed in reviewed documentationNo mention found in the pages reviewed. | Not confirmed in reviewed documentationNo mention found in the pages reviewed. |
| Validation and customer approval of changes | AvailableCode replacements must match 95% of 30+ held-out recorded answers. Model-switch PRs point to an optional replay eval. Every PR is merged by you; gateway optimizations are off until you switch them on.Details: ModelGate docs — code replacements · Details: ModelGate docs — GitHub connection | Not confirmed in reviewed documentationNot applicable as reviewed: no code-change workflow found to validate. | Not confirmed in reviewed documentationNot applicable as reviewed: no code-change workflow found to validate. | Not confirmed in reviewed documentationNot applicable as reviewed: no code-change workflow found to validate. | Not confirmed in reviewed documentationNot applicable as reviewed: no code-change workflow found to validate. |
| Deployment options | LimitedHosted service only. ModelGate OSS is a separate local-first gateway (v0.1) with a smaller feature set.Details: ModelGate OSS | AvailableCloud, open-source gateway, and private-cloud / VPC hosting on Enterprise.Source: Portkey — Pricing (opens in a new tab) | AvailableSelf-hosted open source (MIT); Enterprise adds self-hosted or air-gapped support.Source: LiteLLM docs — Proxy (AI Gateway) (opens in a new tab) · Source: LiteLLM — Enterprise (opens in a new tab) | AvailableCloud, or self-hosted with Docker Compose or Helm; on-premises on Enterprise.Source: Helicone docs — Self-hosting (opens in a new tab) · Source: Helicone — Pricing (opens in a new tab) | AvailableRuns on Cloudflare's network as a managed service.Source: Cloudflare AI Gateway docs (opens in a new tab) |
| Pricing structure | Available$99/month (1 project, +$29 per extra) or $199/month for agencies (5 clients, +$29 per extra), plus 30% of realized savings and 30% of savings from changes made from suggestions for 3 months. 14-day trial.Details: ModelGate pricing | AvailableFree developer tier (10k logs/month); Production $49/month for 100k logs, $9 per extra 100k; Enterprise custom.Source: Portkey — Pricing (opens in a new tab) | AvailableOpen source free; Enterprise pricing not listed publicly.Source: LiteLLM — Enterprise (opens in a new tab) | AvailableFree Hobby tier (10k requests); Pro $79/month and Team $799/month plus usage-based pricing; Enterprise custom.Source: Helicone — Pricing (opens in a new tab) | AvailableCore features free; log storage, Logpush and guardrails billed through Cloudflare Workers / Workers AI usage.Source: Cloudflare AI Gateway — Pricing (opens in a new tab) |
Compare ModelGate with:
Request and cost observability
ModelGateAvailableEvery request logged, failed ones included, and priced against a versioned rate table. Four providers: OpenAI, Anthropic, Google Gemini, Azure OpenAI.Details: ModelGate documentation
PortkeyAvailableLogs, traces, custom metadata and cost attribution across a universal API.Source: Portkey — Pricing (opens in a new tab)
Budgets and access controls
ModelGateAvailablePer-project monthly spend cap (402), requests-per-minute per key (429), token ceilings (413); scoped API keys; agency roles and per-client caps. No SSO.Details: ModelGate security
PortkeyHigher plans onlyRBAC and service-account keys on paid plans. Cost and token budget limits are listed for Enterprise and select Pro users.Source: Portkey — Pricing (opens in a new tab) · Source: Portkey docs — Budget limits (opens in a new tab)
Security guardrails
ModelGateLimitedBuilt-in heuristic detectors: prompt-injection scoring, 17 secret signatures plus an entropy pass, 6 personal-data types. Detect on every plan; enforce on paid plans. No content-moderation categories.Details: ModelGate security
PortkeyNative + integrations20+ built-in deterministic checks and LLM-based checks (including prompt injection), plus partner guardrails such as Aporia and Pillar.Source: Portkey docs — Guardrails (opens in a new tab)
Caching and model routing
ModelGateLimitedExact cache only (temperature 0, no tools, 24 h). Routing moves small, simple calls to a fixed same-provider cheaper model; retry plus one secondary provider. No semantic cache or load balancing.Details: ModelGate documentation
PortkeyAvailableSimple cache on all plans; semantic cache on select Enterprise plans. Load balancing, retries and fallbacks.Source: Portkey docs — Simple and semantic cache (opens in a new tab) · Source: Portkey — Pricing (opens in a new tab)
Repository analysis
ModelGateLimitedGitHub only. Reads the repositories you choose in memory and maps LLM calls to file and line (JavaScript/TypeScript and Python). Source is not stored.Details: ModelGate docs — GitHub connection
PortkeyNot confirmed in reviewed documentationNo mention found in the pages reviewed.
Integration and optimization pull requests
ModelGateLimitedA setup PR that points OpenAI SDK clients at the gateway; one-line PRs for a model switch or a lower output limit; a PR that adds a verified code replacement. Always on a new branch.Details: ModelGate docs — GitHub connection
PortkeyNot confirmed in reviewed documentationNo mention found in the pages reviewed.
Deterministic code replacement
ModelGateLimitedOnly for rule-like tasks (a few fixed labels, extraction, JSON reshaping, arithmetic, simple image checks) with 50+ recorded calls and prompt and response storage on.Details: ModelGate docs — code replacements
PortkeyNot confirmed in reviewed documentationNo mention found in the pages reviewed.
Validation and customer approval of changes
ModelGateAvailableCode replacements must match 95% of 30+ held-out recorded answers. Model-switch PRs point to an optional replay eval. Every PR is merged by you; gateway optimizations are off until you switch them on.Details: ModelGate docs — code replacements · Details: ModelGate docs — GitHub connection
PortkeyNot confirmed in reviewed documentationNot applicable as reviewed: no code-change workflow found to validate.
Deployment options
ModelGateLimitedHosted service only. ModelGate OSS is a separate local-first gateway (v0.1) with a smaller feature set.Details: ModelGate OSS
PortkeyAvailableCloud, open-source gateway, and private-cloud / VPC hosting on Enterprise.Source: Portkey — Pricing (opens in a new tab)
Pricing structure
ModelGateAvailable$99/month (1 project, +$29 per extra) or $199/month for agencies (5 clients, +$29 per extra), plus 30% of realized savings and 30% of savings from changes made from suggestions for 3 months. 14-day trial.Details: ModelGate pricing
PortkeyAvailableFree developer tier (10k logs/month); Production $49/month for 100k logs, $9 per extra 100k; Enterprise custom.Source: Portkey — Pricing (opens in a new tab)
Sources
Official product pages and documentation, reviewed 7 October 2026. Vendors change quickly — check the source before deciding.
- Portkey — Pricing
- Portkey docs — Guardrails
- Portkey docs — Budget limits
- Portkey docs — Simple and semantic cache
- LiteLLM docs — Proxy (AI Gateway)
- LiteLLM docs — Budgets and rate limits
- LiteLLM docs — Guardrails
- LiteLLM docs — Caching
- LiteLLM — Enterprise
- Helicone — Pricing
- Helicone docs — AI Gateway
- Helicone docs — Caching
- Helicone docs — Custom rate limits
- Helicone docs — LLM Security
- Helicone docs — Self-hosting
- Cloudflare AI Gateway docs
- Cloudflare AI Gateway — Dynamic routing
- Cloudflare AI Gateway — Guardrails
- Cloudflare AI Gateway — Pricing
Spotted something out of date? Tell us at support@modelgatehq.com and we’ll correct it. ModelGate’s own claims are explained in the docs, security and pricing pages.