Free LLM Provider Reference — 2026-03-01
Sources: github.com/cheahjs/free-llm-api-resources, github.com/vava-nessa/free-coding-models
Truly Free (No Credits, Ongoing)
| Provider |
Signup URL |
Limits |
Key Env Var |
We Have? |
| OpenRouter |
openrouter.ai |
20 RPM, 50 req/day |
OPENROUTER_API_KEY |
✅ |
| Google AI Studio |
aistudio.google.dev |
5 RPM, 20 req/day (Flash) |
GEMINI_API_KEY |
✅ |
| NVIDIA NIM |
build.nvidia.com |
40 RPM (phone verify) |
NVIDIA_API_KEY |
❌ |
| Mistral |
console.mistral.ai |
1 RPS, 500K TPM, 1B tokens/month |
MISTRAL_API_KEY |
✅ |
| Codestral |
console.mistral.ai |
30 RPM, 2000 req/day |
CODESTRAL_API_KEY |
❌ |
| HuggingFace |
huggingface.co |
$0.10/month credits |
HF_TOKEN |
❌ |
| Vercel AI Gateway |
vercel.com |
$5/month |
AI_GATEWAY_API_KEY |
❌ |
| Cerebras |
cloud.cerebras.ai |
30 RPM, 60K TPM, 1M TPH, 1M TPD |
CEREBRAS_API_KEY |
❌ |
| Groq |
console.groq.com |
Per-model (see detail) |
GROQ_API_KEY |
✅ |
| Cohere |
dashboard.cohere.com |
20 RPM, 1000 req/month |
COHERE_API_KEY |
❌ |
| GitHub Models |
github.com/marketplace/models |
Per Copilot tier |
GITHUB_TOKEN |
❌ |
| Cloudflare Workers AI |
dash.cloudflare.com |
10K neurons/day |
CLOUDFLARE_API_TOKEN |
❌ (have DNS key) |
| SambaNova |
cloud.sambanova.ai |
$5/mo + $30/mo w/ payment |
SAMBANOVA_API_KEY |
❌ |
Trial Credits (One-Time)
| Provider |
Credits |
Duration |
Signup URL |
Key Env Var |
We Have? |
| Fireworks |
$1 |
? |
fireworks.ai |
FIREWORKS_API_KEY |
❌ |
| Baseten |
$30 |
? |
baseten.co |
- |
❌ |
| Nebius |
$1 |
? |
studio.nebius.ai |
- |
❌ |
| Novita |
$0.50 |
1 year |
novita.ai |
- |
❌ |
| AI21 |
$10 |
3 months |
studio.ai21.com |
- |
❌ |
| Upstage |
$10 |
3 months |
upstage.ai |
- |
❌ |
| NLP Cloud |
$15 |
? |
nlpcloud.com |
- |
❌ |
| Alibaba Intl |
1M tokens/model |
? |
alibabacloud.com |
ALIBABA_CLOUD_API |
✅ (broken) |
| Modal |
$5/mo free + $30/mo w/ CC |
? |
modal.com |
- |
❌ |
| Inference.net |
$1 (+$25 survey) |
? |
inference.net |
- |
❌ |
| Hyperbolic |
$1M free tokens |
? |
hyperbolic.xyz |
HYPERBOLIC_API_KEY |
❌ |
| Scaleway |
$5 |
3 months |
scaleway.com |
SCALEWAY_API_KEY |
❌ |
Priority Signups (highest value-to-effort)
- NVIDIA NIM — 40+ S+/S tier models, all free. Phone verify only.
- Cerebras — Fastest inference available. GLM, Llama. Free.
- SambaNova — Fast inference. Free tier generous.
- Codestral — Free separate key from Mistral for coding. 30 RPM.
- HuggingFace — Huge model catalog. Tiny free tier but useful.
- Cohere — Command R+ is excellent. 1000 req/month free.
- Cloudflare Workers AI — We already have CF account! Just need Workers AI token.
Groq Detailed Limits (Free Tier)
| Model |
RPD |
TPM |
| Llama 3.3 70B |
1,000 |
12,000 |
| Llama 4 Maverick |
1,000 |
6,000 |
| Kimi K2 Instruct |
1,000 |
10,000 |
| GPT-OSS-120B |
1,000 |
8,000 |
| Qwen3 32B |
1,000 |
6,000 |
| Whisper Large v3 |
2,000 |
7,200 audio-sec/min |
Cerebras Detailed Limits (Free Tier)
| Model |
RPM |
TPM |
TPH |
TPD |
| GPT-OSS-120B |
30 |
60,000 |
1,000,000 |
1,000,000 |
| Qwen3 235B |
30 |
60,000 |
1,000,000 |
1,000,000 |
| Llama 3.3 70B |
30 |
64,000 |
1,000,000 |
1,000,000 |
| GLM-4.6 |
10 |
60,000 |
100,000 |
1,000,000 |