Free AI models
Free tiers you can sign up for directly with each provider, with their real limits. Free limits change often, so each row says how sure we are.
| Provider | Model | Free limit | Token limit | Context | Tool calls | Confidence |
|---|---|---|---|---|---|---|
| Groq | gpt-oss-120b | 1,000 requests/day, 30/min (free plan, gpt-oss-120b) | 8,000 tokens/min (checked live) and about 200,000 tokens/day | 131k | Yes | reported |
| Google Gemini | Gemini 3.5 Flash-Lite | Google shows your live limit in AI Studio; Flash-Lite free tier reported around 500/day | about 250,000 tokens/min reported | 1000k | Yes | reported |
| Cloudflare Workers AI | gpt-oss-120b | 10,000 Neurons/day free; about 150 chat turns | about 300,000 tokens/day (10,000 Neurons) | 128k | Yes | estimate |
| Mistral | Mistral Small | Free plan: about $10 of API credit a month (since Aug 2026), shared with Mistral Studio | – | 128k | Yes | estimate |
| Z.ai (GLM) | GLM-4.5-Flash | Free model; no published daily cap (limited by concurrent requests). Dev1 AI stops at 2,000 turns/day and backs off on a real 429. | – | 128k | Yes | estimate |
| NVIDIA NIM | Nemotron 3 Ultra | About 40 requests/min per key; no published daily number. Dev1 AI stops at 1,000 turns/day and backs off on a real 429. | – | 128k | Yes | estimate |
| OpenRouter (free models) | Rotating free models | 50 requests/day on free models (1,000/day after a one-time $10 credit purchase) | free models vary; Dev1 AI assumes 32k context | 32k | No | reported |
Confidence: “estimate” = the vendor publishes no number; this is an estimate; “reported” = the limit comes from third-party reports of the vendor's limits. These are separate accounts, not part of AnyModl. When you outgrow a free tier, AnyModl gives you one prepaid key for paid models, no subscription. See pricing.