The Gemini API is the only one of these two with a standing free tier. Flash and Flash-Lite models, embeddings and most audio models run at no charge inside the free rate limits, and Google uses that traffic to improve its products. OpenAI has no free allowance; its Build tier opens once you’ve bought $5 of credits. For a prototype that’s the whole decision. Everything below was checked on both companies’ own pages on November 2, 2026.
Past the prototype the gap narrows. GPT-6.1 Sol costs $2 per million input tokens and $10 output, and Gemini 4 Argon is the same today but doubles to $4 and $20 on December 1, 2026. Both platforms now run on prepaid credits, both discount batch work by 50%, and both cap monthly spend by tier. The founder question is how far free carries you and what the paid tier looks like after.
What free gets you and what it costs
Gemini 3.8 Flash, 3.6 Flash, 3.5 Flash-Lite, 3.1 Flash-Lite, 2.5 Flash-Lite, the Live and TTS models and Gemini Embedding 2 are all available on the free tier. Gemini 4 Argon, 3.1 Pro Preview and the image models are paid only. Per-model free limits aren’t published as a table anymore; each model’s RPM, TPM and RPD show on the AI Studio rate limit screen, limits apply per project, and daily counts reset at midnight Pacific. Keep your demo project separate from the one your app uses.
The price of free is your data. Google uses content submitted to the unpaid services, and the responses, to provide, improve and develop its products, and human reviewers may read, annotate and process that input and output after it’s disconnected from your account. Google tells you plainly not to submit sensitive or confidential information on the unpaid tier. Paid traffic isn’t used to improve products.
OpenAI has no equivalent. Its rate limit docs list a Free tier, but that is a $100 monthly usage cap for organizations in an allowed geography, and the only no-cost items are a free moderation model and 1 GB of file search storage. Five dollars gets you in the door, and that’s the cheapest way to test.
Paid prices and the increases already scheduled
Today the flagships match. Gemini 4 Argon and GPT-6.1 Sol both list at $2 in and $10 out. The difference is the calendar. Argon rises to $4 and $20 on December 1, 2026, and Gemini 3.8 Flash and 3.6 Flash rise from $0.75 and $3.75 to $1.50 and $7.50 in 2027. A budget built on Gemini list prices in October will be wrong by January. Plan on the higher figure if your launch is later than that.
At the cheap end Gemini still wins. Gemini 2.5 Flash-Lite is $0.10 in and $0.40 out, and Gemini 3.1 Flash-Lite is $0.25 and $1.50. OpenAI’s GPT-6 Luna is $0.10 and $0.50, with cached input at $0.01. Gemini context caching on 3.8 Flash costs $0.075 per million cached tokens plus $0.50 per million tokens per hour of storage, so a cache you keep warm all day isn’t free the way OpenAI’s automatic cache is.
Both discount batch by half. Gemini batch rates are listed per model, for example $1 in and $5 out on Argon. OpenAI’s Batch API returns within 24 hours and has its own queue limits, and its Flex mode bills synchronous-looking calls at batch rates with slower responses and occasional 429s that aren’t charged.
How much you can spend each month
Gemini tiers hang off a Cloud Billing account. Tier 1 arrives the moment you link billing and carries a $250 monthly cap, Tier 2 needs $100 paid plus 3 days and raises the cap to $2,000, and Tier 3 needs $1,000 paid plus 30 days for $20,000 to $100,000 and up. There’s also a rolling 10 minute spend limit, $10 on Tier 1 and $200 on Tiers 2 and 3, that returns a 429 when crossed. A traffic spike on Tier 1 hits that wall fast.
OpenAI’s three paid tiers are simpler and larger. Build at $5 of purchases allows $500 a month, Launch at $100 allows $5,000, and Grow at $500 allows $200,000. On Build, the Sol model family gets 5,000 requests and 1 million tokens per minute, and Luna gets 2 million tokens per minute. For a team that will spend a few thousand dollars a month, OpenAI reaches the needed cap with less money and no waiting period.
Prepaid billing and the trial loophole that closed
Both are prepaid now. On Gemini, Prepay is the default for new users, with a $5 minimum, a $5,000 maximum per purchase, credits that expire after 12 months and a 402 error when the balance hits zero. Postpay is being phased out for Gemini API usage. OpenAI’s tiers are measured in total credit purchases, so an account can’t run up a bill it hasn’t funded. Neither will surprise you with an invoice.
One trap worth knowing: the $300 Google Cloud welcome credit no longer pays for Gemini API or AI Studio usage. Gemini API has been excluded from the free trial since March 2026, and only welcome credits granted before that change can still be spent on it. Don’t plan a Gemini build around the Cloud trial.
Startup credits and who qualifies
Google’s route is the Google for Startups Cloud Program. It gives 100% credit on Google Cloud spend up to $250,000 in year one for AI startups, then a 20% credit up to $100,000 in year two, and those credits cover Gemini and Gemma. Eligibility is Pre-Seed to Series A, founded within the last three to five years, a working MVP, a verified company domain and no prior Google Cloud credits above $5,000.
OpenAI for Startups needs a referral code from a partner VC and publishes no credit figure. Free API credits, rate limit upgrades and solutions engineer time are the possible benefits, and billing reverts to standard once credits are spent. If your investor isn’t a partner, there’s no self-serve application to fall back on. That’s a hard wall for a bootstrapped team.
Where each one fits
Gemini is the better sandbox and the better deal for a funded AI startup that qualifies for Cloud credits. OpenAI is the steadier production bill because its list prices have no announced increases and its $500 purchase opens a $200,000 monthly cap. Plenty of teams prototype on free Gemini Flash and move paid traffic wherever the model fits, which both billing systems allow without a contract. We’d do the same.
The winner: Gemini API
A free tier that covers current Flash models, a $0.10 Flash-Lite, and a startup program worth up to $350,000 outweigh OpenAI’s simpler tiers for most readers of this page. The caveats are real: free traffic trains Google’s products, and Argon doubles in price on December 1, 2026, so paid Gemini is no bargain next year.
Choose the Gemini API if you want free daily usage for prototypes, can accept product improvement use of free traffic, or qualify for Google Cloud startup credits.
Gemini API offers (4)Choose OpenAI if you need stable pricing without scheduled increases, a $200,000 monthly cap after $500 spent, and no data sharing on any tier.
OpenAI API offers (5)