Setup
Free API tiers and credits for getting started with BYOK
Legitimate ways to try BYOK apps without paying: Google AI Studio's free tier, Groq's free tier, OpenRouter's free models and local models, with the limits and data terms to check first.
By the BYOK Tools editorial team · 22 November 2026 · 6 min read
You can test most BYOK apps without spending money. Several providers offer free tiers or free models, and local models cost nothing per token. Free access comes with lower rate limits and, in some cases, different data terms. Limits change often, so check each provider's current documentation before you rely on it.
Google AI Studio
Google AI Studio issues Gemini API keys with a free tier on selected models. It is rate-limited by requests per minute and per day, and the models available on the free tier change over time.
Read the data terms. Google's terms for unpaid usage can permit use of prompts and responses to improve its products, which differs from paid usage. Do not send confidential material through the free tier. Enabling billing on the linked Google Cloud project moves the key to paid terms with higher limits.
Google also provides an OpenAI-compatible endpoint for Gemini, so the free key works in many apps that only support OpenAI-style providers. See OpenAI-compatible endpoints.
Groq
Groq runs open-weight models on its own inference hardware and is known for fast responses. Its free tier gives API access with per-model rate limits on requests and tokens. The API is OpenAI-compatible at https://api.groq.com/openai/v1, so most BYOK apps can use it through a custom provider setting.
OpenRouter free models
OpenRouter lists some models at no cost. Their IDs end in :free. Free models have tighter rate limits, and limits can depend on whether your account holds purchased credit. Availability changes as hosts add and remove free capacity. Some free endpoints are offered by hosts that may log prompts, so check the model's provider details and your OpenRouter privacy settings. See OpenRouter explained.
Local models
Running a model locally through Ollama, LM Studio or Jan has no per-token cost and no external rate limits. Data stays on your machine. The trade-off is hardware: quality and speed depend on available memory and GPU. See local AI with Ollama.
New-account credits
Some providers grant trial credits to new accounts, sometimes tied to phone verification or a promotion. Offers come and go, so check each provider's billing page rather than relying on third-party lists.
Practical advice
- Treat free keys like paid keys. A leaked free key can still be abused, and it can start incurring charges once billing is enabled on the account. Follow the security checklist.
- Expect HTTP 429 errors. Free tiers hit rate limits quickly, especially with coding agents that make many requests. Choose apps that retry with backoff.
- Check data terms per tier. Free and paid tiers from the same provider can have different retention and training policies.
- Move to paid when it matters. For sensitive data, team use or consistent throughput, add a small prepaid balance and a spend limit. Our cost calculator estimates what typical usage costs.
Editorial independence: rankings are based on our own testing and research. Some outbound links may in future be affiliate links; that never affects scores. See our editorial policy.
Get the BYOK Brief
New listings, price changes and model releases. Fortnightly.