Gateways, routers, and direct model APIs ranked free tier first, then paid. Free access types are labeled honestly — always-free, monthly credit, signup credit, or trial. Every row links an official source and a real verification date. We do not invent free tiers.
Free input & output tokens on multiple Gemini models via AI Studio
Free
Free tier: free input and output tokens on multiple models, free context caching, free AI Studio access. Rate-limited; some data used to improve Google products.
Paid pricing
Paid tiers after free allotment; Flash-class intro pricing $0.75 / $3.75 per 1M input/output through Dec 31, 2026 (ai.google.dev).
Gemini family (limited model set on free tier)frontier free usageprototypinglong context
Caveat: Free tier rate-limited; direct page fetch intermittent — verify limits in AI Studio console.
10,000 Neurons/day free on every Workers plan — no credit card
Free
10,000 Neurons per day free (resets 00:00 UTC) on Free and Paid Workers plans. Beyond free: $0.011 / 1,000 Neurons. Some frontier models require a paid billing method.
Free $0 plan with monthly free-model allowance — no card required
Free
Free plan $0/month: free monthly allowance (4-week cycle), rotating free models (e.g. DeepSeek V4.1 Flash, MiMo V2.6 Flash, TH-Rudder), one OpenAI-compatible API, no card. Free models must be enabled in dashboard; free routes never charge balance.
Paid pricing
Pay-as-you-go wallet (top-ups from $10) + optional passes: Agent $0.99 first month then $1.99/mo ($10 usage); Office $9.99/mo ($35); Frontier $99/mo ($180). 0% token markup currently.
9+ premium models + free set (official marketing; exact catalog size not hard-counted)free allowanceagent APIone wallet
Caveat: Some marketing pages mention welcome credit; official docs say accounts start at $0 with no welcome credit.
Free gateway: 1,000 requests/month, 2 of 13 providers, BYOK, zero markup
Free
Free plan $0 flat forever: 1 seat, 1 team, 2 connected providers, 1,000 gateway requests/month, budgets/hard caps/analytics, no credit card.
Paid pricing
Starter $29/mo (1M requests, all 13 providers); Team $99/mo (10M); Enterprise $499/mo — flat subscription, zero token markup (providers bill at exact rates).
Free plan ($0) documented for getting started with APIs; community support; downgrade path from Developer confirmed in billing FAQ. Exact free-tier RPM/TPM not published on public docs — check console.
Paid pricing
Paid tiers with higher rate limits (Developer and above; see GroqCloud billing).
Open and partner models on Groq LPU inferencespeedfree API entrylow latency
Caveat: Do not publish specific free-tier RPM/TPM without reading console — not on public rate-limits page.
Free tier with monthly free credits on a model subset — zero token markup
Free
Every Vercel team gets a free tier: monthly free credit (amount not restated on the fetched pricing body — check dashboard), free-tier model subset, lower rate limits. Free credits start on first AI Gateway request; buying credits moves you to paid tier.
Paid pricing
Zero markup on tokens at provider list price; BYOK no fee. Paid AI Gateway Credits; volume discounts available.
Multi-provider catalog (free subset only on free tier)Vercel appszero markupfree model subset
Caveat: Free credit amount not printed on the pricing page body we loaded — verify in Vercel dashboard before quoting a $ figure.
$150 free credits via official claim form (not instant — form-based). Docs historically stated no free trial; current official page offers this claim path.
How this list works: we load each provider's official pricing or docs page, record only claims printed there, and date the check. If a free tier could not be verified on an official page, the provider is listed as paid or omitted — not softened into "free".