Free AI API tiers in 2026: what Gemini, OpenAI, DeepSeek and Mistral give for free
Which AI APIs can you use for free? Google's Gemini API has a free tier; OpenAI, DeepSeek and Mistral list none for text models. The details, and the cheapest paid options.
Of the four providers checked, only Google offers a free tier for its text models: the Gemini API’s free tier covers Gemini 3.8 Flash, 3.5 Flash-Lite and several older models, but not the flagship Gemini 4 Argon. OpenAI, DeepSeek and Mistral list no free token allowance for text models on their pricing pages. If you need near-free rather than free, the cheapest paid models cost $0.10 per million input tokens.
Provider by provider
| Provider | Free tier for text models? | What the pricing page says |
|---|---|---|
| Google (Gemini API) | Yes | “Free of charge” listed for Gemini 3.8 Flash, 3.6 Flash, 3.5 Flash-Lite, 3.1 Flash-Lite, 2.5 Pro, 2.5 Flash, 2.5 Flash-Lite and others. Not available for Gemini 4 Argon or Gemini 3.1 Pro Preview. |
| OpenAI | No | Only the omni-moderation-latest model is priced “Free”. File search and ChatKit storage include 1 GB free. |
| DeepSeek | No amount stated | No free grant is stated; the page says any granted balance is used before topped-up balance. |
| Mistral | No | No free API tier on either pricing page; the free plan covers limited access to Vibe and Mistral Studio. |
The Gemini free tier, in practice
The free tier is the easiest way to prototype with a capable model at no cost. Two things to know before you build on it:
- The limits are not on the public page. Google’s rate-limits page does not publish the free tier’s requests per minute, tokens per minute or requests per day; it points you to AI Studio to view the limits active on your own account. Check there before you plan traffic.
- The flagship is paid only. Gemini 4 Argon is not on the free tier. On paid usage it costs $2 per million input tokens and $10 per million output tokens until Google’s announced increase to $4 and $20 on December 1, 2026.
Google’s pricing page also announces that Gemini 3.8 Flash’s paid price will rise from $0.75 input and $3.75 output to $1.50 and $7.50 on January 1, 2027. If you start free and plan to scale, budget with the new prices.
The cheapest paid options
When a free tier’s limits are too tight, these models cost the least per million tokens at standard rates:
| Model | Input | Output |
|---|---|---|
| GPT-6 Luna (OpenAI) | $0.10 | $0.50 |
| Claude Haiku 5.5 (Anthropic, prompts up to 100,000 tokens) | $0.10 | $0.50 |
| Mistral Small 4 | $0.15 | $0.60 |
| DeepSeek-V4.1-Flash (peak; off-peak is half) | $0.30 | $1.20 |
| Gemini 3.5 Flash-Lite (paid tier) | $0.30 | $2.50 |
At $0.10 input and $0.50 output, a million short requests of 500 tokens in and 200 out would cost $50 + $100 = $150. Google’s page also still lists older models at lower prices, such as Gemini 2.5 Flash-Lite at $0.10 input and $0.40 output.
DeepSeek’s off-peak discount deserves a note: its off-peak rates are half the peak rates, so batch jobs that can run outside 01:00 to 04:00 and 06:00 to 10:00 UTC on weekdays cost half as much.
Estimate before you build
The API cost calculator prices your expected traffic on every model at once, so you can see what leaving the free tier would cost.
Sources
- Google, Gemini API pricing and rate limits.
- OpenAI, API pricing.
- DeepSeek, models and pricing.
- Mistral, pricing and API pricing.
- Anthropic, API pricing.
All pages read on October 26, 2026.
Free toolWhat will 10,000 chats a month cost?Estimate my bill
