Is Gemini 3.1 Pro or GPT-4o cheaper?

For a typical request (10,000 input + 2,000 output tokens), Gemini 3.1 Pro is cheaper — about 2% less, or roughly $100 saved per 100,000 requests. Gemini 3.1 Pro runs $2/$12 per 1M input/output tokens; GPT-4o runs $2.5/$10.

Which has the larger context window?

Gemini 3.1 Pro, at 1,000,000 tokens versus 128,000.

How accurate are these token counts?

Gemini 3.1 Pro: Approximated with o200k_base; drift typically ~3% on English and code. GPT-4o: Exact tokenization via the canonical OpenAI vocab (o200k_base). The dollar math itself is exact once the token count is known.

Gemini 3.1 Pro vs GPT-4o: pricing & cost comparison

On input tokens, Gemini 3.1 Pro is the cheaper of the two — 20% less per million ($2 vs $2.5). On output, GPT-4o is 17% cheaper ($12 vs $10) — and since output is usually the dominant cost driver, that gap matters more than it looks.

Side by side

	Gemini 3.1 Pro	GPT-4o
Input / 1M tokens	$2	$2.5
Output / 1M tokens	$12	$10
Context window	1,000,000	128,000
Token-count accuracy	±3%	exact
Cost — 10,000 input + 2,000 output tokens	$0.044	$0.045

What a real request costs

Take a representative turn — 10,000 input + 2,000 output tokens. Gemini 3.1 Pro comes to $0.044, GPT-4o to $0.045. Across 100,000 requests that's a $100 swing in favour of Gemini 3.1 Pro. To run the numbers on your actual prompt, paste it into the calculator and toggle Compare across all models.

Close call — decide on output cost

Gemini 3.1 Pro and GPT-4o are close enough on price that the tie-breakers matter more than the headline rate. Output tokens are usually the dominant cost driver, so weigh $12 vs $10 per 1M output first, then context window — Gemini 3.1 Pro carries the larger one at 1,000,000 tokens. They're different vendors, so factor in a small tokenizer calibration buffer on whichever side isn't exact.

See the full breakdown on the dedicated pages for Gemini 3.1 Pro and GPT-4o.

FAQ

Is Gemini 3.1 Pro or GPT-4o cheaper?: For a typical request (10,000 input + 2,000 output tokens), Gemini 3.1 Pro is cheaper — about 2% less, or roughly $100 saved per 100,000 requests. Gemini 3.1 Pro runs $2/$12 per 1M input/output tokens; GPT-4o runs $2.5/$10.
Which has the larger context window?: Gemini 3.1 Pro, at 1,000,000 tokens versus 128,000.
How accurate are these token counts?: Gemini 3.1 Pro: Approximated with o200k_base; drift typically ~3% on English and code. GPT-4o: Exact tokenization via the canonical OpenAI vocab (o200k_base). The dollar math itself is exact once the token count is known.

Both prices are computed from tokenmath's verified pricing table. Rates sourced from ai.google.dev and openai.com, verified 2026-07-06. Vendor pricing changes often — confirm before you commit.