Home / Comparisons / Best free LLM APIs, no card
Best free LLM APIs with no credit card, compared
Short answer: six verified LLM APIs don’t require a credit card — Gemini, Groq, OpenRouter, Cloudflare, Zhipu GLM Flash, and SiliconFlow. The only long-term free one is GLM Flash; the rest are daily allowances or one-time credits. None is unlimited.
Comparison table
| Offer | Free allowance | Requirements | Reset | Verified |
|---|---|---|---|---|
| Gemini API free tier | Free tier for Gemini Flash and other models, 5–30 RPM / per-model daily caps | Account | Resets daily; actual limits as shown in the AI Studio quota panel | 2026-11-08 |
| Free Plan fast inference | Per model: 30 RPM / 1,000 requests per day / 200K tokens per day | Account | Per-model rate limits; resets daily | 2026-11-08 |
| :free free models | 20 RPM / 50 requests per day; 1,000 per day after the account has been topped up $10 | Account | Resets daily; model IDs end in :free | 2026-10-27 |
| Daily free Neurons | 10,000 Neurons per day | Account | Resets at 00:00 UTC (08:00 Beijing time) | 2026-11-01 |
| GLM-4.5-Flash / GLM-4.7-Flash free models | Officially marked fully free; concurrency shown per account | Account | Long-term free models; no claim needed; call right after registering | 2026-11-08 |
| Signup credits + real-name verification voucher | ¥14 credit on signup; invite + real-name verification earns a ¥16 voucher (180 days) | Account, Real-name verification | One-time credit; voucher valid 180 days | 2026-10-27 |
How to choose
- Long-term free, no claiming: Zhipu GLM Flash — officially marked fully free, callable right after registering; only the Flash series is free, flagship models are paid.
- Most models through one key: OpenRouter :free models — 20 RPM / 50 requests per day, rising to 1,000 per day after the account has been topped up $10. Note that failed requests also count toward the daily cap.
- Speed: Groq Free Plan — per-model limits of 30 RPM / 1,000 requests per day / 200K tokens per day; limits are calculated per organization and per model.
- Long context: Gemini API free tier — Flash / Flash-Lite models with context up to 1M; limits are 5–30 RPM with per-model daily caps, and actual limits are shown in the AI Studio quota panel.
- Simplest daily allowance: Cloudflare Workers AI — 10,000 Neurons per day, resetting at 00:00 UTC (08:00 Beijing time); some newer models require a payment method first.
- Credit-style balance: SiliconFlow — ¥14 credit on signup, plus a ¥16 voucher (180 days) for invite + real-name verification; real-name verification is required.
All six are API access and OpenAI-compatible. None is unlimited: each has rate limits, a daily cap, or a credit balance. For the full picture of how free allowances differ, see the free-tier types comparison.
FAQ
Which of the six is truly long-term free?
Zhipu GLM Flash: the verified listing says it is officially marked fully free, with concurrency shown per account, cadence ongoing, and callable right after registering. Verified 2026-11-08 — check the official page before claiming.
Do any of the six require a card?
None of the six lists card binding in its verified requirements. Two caveats: Cloudflare Workers AI notes that some newer models require a payment method first, and SiliconFlow requires real-name verification. Check the official page before claiming.
Disclaimer: table data comes from vendor official pages; allowances can change or end at any time — check the official page before claiming.