The free AI API tracker. Every limit, dated.
Free quotas, rate limits, card requirements and trial credits for the APIs indie developers, students and agent builders actually use — each number dated and linked to the provider's own docs.
8 frontier models now require a paid billing method
23 providers. One honest table.
Blank means the provider doesn't publish it, or we couldn't verify it — we never fill gaps with guesses. Click a row for details and sources.
| Best free model | Free allowance | Expand | |||
|---|---|---|---|---|---|
| Chat & code LLMs15 | |||||
Cloudflare Workers AIFree tier · LLM + multimodal inference |
llama-3.3-70b-instruct | 10,000 Neurons/day | Unverified | Oct 21 | |
10,000 Neurons per day free on both Workers Free and Paid; some frontier models now require a paid billing method. Official sourcesVerified Oct 21, 2026Full Cloudflare Workers AI report | |||||
CohereFree tier · LLM + embeddings + rerank |
Command A | 1,000 calls/month | Unverified | Oct 21 | |
Free trial keys: 1,000 API calls per month, 20 req/min on Chat models, plus Embed and Rerank. 20req / min Official sourcesVerified Oct 21, 2026Full Cohere report | |||||
Google Gemini APIFree tier · LLM inference |
gemini-3.8-flash | Free tokens; limits in AI Studio | Not required | Oct 21 | |
Free tier on Flash, Flash-Lite, 2.5 Pro, Gemma 4 and embeddings; newest flagship (Gemini 4 Argon) is paid-only. Official sourcesVerified Oct 21, 2026Full Google Gemini API report | |||||
GroqFree tier · LLM inference |
openai/gpt-oss-120b | 1,000 req/day | Not required | Oct 21 | |
Permanent Free plan with per-model daily caps on open-weight models (gpt-oss, Qwen, Whisper). 30req / min 1Kreq / day 8Ktokens / min Verified Oct 21, 2026Full Groq report | |||||
Mistral AIFree tier · LLM inference |
— | Monthly allowance (unpublished) | Unverified | Oct 21 | |
'Free mode' includes a monthly usage allowance; exact limits are only shown in your Admin Panel. Official sourcesVerified Oct 21, 2026Full Mistral AI report | |||||
NVIDIA NIMFree tier · LLM + multimodal inference |
API catalog models | Free for prototyping | Unverified | Oct 21 | |
Free NIM API endpoints for prototyping via the NVIDIA Developer Program; production use needs an AI Enterprise license. Verified Oct 21, 2026Full NVIDIA NIM report | |||||
Ollama CloudFree tier · LLM inference |
Starter models | Starter usage/month | Unverified | Oct 21 | |
Free plan includes a starter amount of cloud usage on starter models, 1 concurrent request; local models are always unlimited. Official sourcesVerified Oct 21, 2026Full Ollama Cloud report | |||||
SambaNova CloudFree tier · LLM inference |
DeepSeek-V3.1 | 20 req/day per model | Unverified | Oct 21 | |
Free Tier per model: 20 requests/min, 20 requests/day, 200K tokens/day on DeepSeek, Llama 3.3 70B, gpt-oss-120b, Gemma 4. 20req / min 20req / day Official sourcesVerified Oct 21, 2026Full SambaNova Cloud report | |||||
Scaleway Generative APIsFree tier · LLM inference (EU) |
gpt-oss-120b | 1M tokens, one-time | Required | Oct 21 | |
First 1,000,000 tokens and 60 minutes of audio transcription free; EU-hosted, payment method required. 300req / min 200Ktokens / min Official sourcesVerified Oct 21, 2026Full Scaleway Generative APIs report | |||||
Z.aiFree tier · LLM inference |
GLM-4.7-Flash | Free per-token price | Unverified | Oct 21 | |
GLM-4.7-Flash, GLM-4.5-Flash and GLM-4.6V-Flash are priced 'Free' for input and output. Official sourcesVerified Oct 21, 2026Full Z.ai report | |||||
Cerebras InferenceTrial credits only · LLM inference |
gpt-oss-120b | $5 trial · 30 days | Required | Oct 21 | |
No permanent free tier: $5 trial credits after adding a verified payment method, expiring after 30 days. 5req / min 30Ktokens / min Official sourcesVerified Oct 21, 2026Full Cerebras Inference report | |||||
Fireworks AITrial credits only · LLM inference |
— | $1 credit · 10 req/min | Not required | Oct 21 | |
$1 starter credit usable without a payment method, capped at 10 requests/min account-wide. 10req / min Verified Oct 21, 2026Full Fireworks AI report | |||||
DeepInfraNo free tier · LLM inference |
— | No free credits | Required | Oct 21 | |
No sign-up credit: 'You have to add a card or pre-pay or you won't be able to use our services.' Official sourcesVerified Oct 21, 2026Full DeepInfra report | |||||
Together AINo free tier · LLM inference |
— | No free trial · $15 minimum | Required | Oct 21 | |
No free trial: fully prepaid with a $15 minimum first credit purchase. Verified Oct 21, 2026Full Together AI report | |||||
GitHub ModelsRetired · LLM inference |
— | Retired Jul 30, 2026 | — | Oct 21 | |
Fully retired on July 30, 2026 — playground, catalog, inference API and BYOK are gone. Official sourcesVerified Oct 21, 2026Full GitHub Models report | |||||
| Gateways & routers3 | |||||
OpenRouterFree tier · Gateway / router |
nemotron-3-super-120b:free | 50 req/day per account · 1,000 with ≥10 credits | Not required | Oct 21 | |
17 ':free' model variants today; 50 free requests/day per account, or 1,000/day once you have bought ≥10 credits. 20req / min 50req / day Verified Oct 21, 2026Full OpenRouter report | |||||
Vercel AI GatewayFree tier · Gateway / router |
Free-tier models | $5 credit/month | Unverified | Oct 21 | |
$5/month of AI Gateway credit on every Vercel team, for a subset of 'Free Tier eligible' models. Verified Oct 21, 2026Full Vercel AI Gateway report | |||||
Hugging Face Inference ProvidersNo free tier · Gateway / router |
— | No free credits | Required | Oct 21 | |
Free users get no monthly inference credits; PRO users and Team/Enterprise seats get $2.00/month. Official sourcesVerified Oct 21, 2026Full Hugging Face Inference Providers report | |||||
| Embeddings & search2 | |||||
Jina AIFree tier · Embeddings + web reader |
Embeddings, Reader | 100 req/min (free key) | Unverified | Oct 21 | |
Reader works without a key (20 RPM); a free API key raises Reader to 500 RPM and unlocks Embeddings/Rerank at 100 RPM. 100req / min 100Ktokens / min Official sourcesVerified Oct 21, 2026Full Jina AI report | |||||
Voyage AIFree tier · Embeddings + rerank |
voyage-4 | 200M tokens per model | Unverified | Oct 21 | |
First 200M tokens free per account on voyage-4 family & rerank-3; 50M on older domain models. 2Kreq / min 8Mtokens / min Official sourcesVerified Oct 21, 2026Full Voyage AI report | |||||
| Speech & audio3 | |||||
AssemblyAIFree tier · Speech-to-text |
Universal-3.5 Pro | 185 h transcription | Not required | Oct 21 | |
No credit card: up to 185 hours of pre-recorded and 333 hours of streaming transcription free. Official sourcesVerified Oct 21, 2026Full AssemblyAI report | |||||
ElevenLabsFree tier · Text-to-speech / audio |
Text to Speech | 10k credits/month | Unverified | Oct 21 | |
Free plan: 10k credits per month across TTS, STT, sound effects and voice design. Official sourcesVerified Oct 21, 2026Full ElevenLabs report | |||||
DeepgramTrial credits only · Speech-to-text / TTS |
Speech-to-text | $200 credit | Not required | Oct 21 | |
$200 free credit, then pay-as-you-go; no credit card required. Official sourcesVerified Oct 21, 2026Full Deepgram report | |||||
What changed. Dated, with a source.
No number without a source.
Most free-tier lists copy each other. We read the provider's own page, write down the date, and say so when a value can't be verified.
- Official pages only.Docs, pricing pages and public APIs — never third-party lists.
- Blanks stay blank.Unpublished limits are marked unverified instead of estimated.
- Every value is dated.Each provider shows when it was last checked against its sources.
- Changes are diffed daily.A script re-reads every source and flags what moved for human review.
The Monday diff.
One email a week with every free-tier change we caught — what was cut, what became free, and the source for each. Nothing else.
Subscriptions aren't open yet — this form doesn't send or store anything.
Weekly, free, unsubscribe anytime · Privacy