THE TICKER · GOOGLE · CHECKED DAILY · LAST CHANGE 2026-07-11
GEMINI 3 FLASH API PRICING
What Google charges for Gemini 3 Flash, in USD, verified daily against public list prices. Effective price — if there's a promo running, this is the promo.
INPUT / 1M TOKENS
$0.500
OUTPUT / 1M TOKENS
$3.00
INPUT / 1K TOKENS
$0.000500
OUTPUT MULTIPLE
6.0x
output vs. input price
WHAT THINGS ACTUALLY COST ON GEMINI 3 FLASH
| Scenario | Tokens in / out | 1 run | 1,000 runs |
|---|---|---|---|
| A typical prompt (300 tokens in) | 300 / ~900 | $0.0029 | $2.85 |
| A 20-turn conversation (context re-sent) | 4,000 / ~4,000 | $0.01 | $14.00 |
| Summarizing a 10-page document | 8,000 / ~3,200 | $0.01 | $13.60 |
| One agent run (tools, retries, thinking) | 2,000 / ~12,000 | $0.04 | $37.00 |
| A code review request | 1,500 / ~3,750 | $0.01 | $12.00 |
Output sizes assume the same editorial ratios as the Burn Calculator (3x for prompts, 6x for agent runs). Estimates, not invoices.
VS. THE REST OF THE BOARD
| Model | Input / MTok | Output / MTok | Same job costs |
|---|---|---|---|
| Claude Fable 5 Anthropic → | $10.00 | $50.00 | 16.8x more |
| Claude Opus 4.8 Anthropic → | $5.00 | $25.00 | 8.4x more |
| GPT-5.6 OpenAI → | $5.00 | $30.00 | 10.0x more |
| GPT-5.5 OpenAI → | $5.00 | $30.00 | 10.0x more |
| GPT-5.4 OpenAI → | $2.50 | $15.00 | 5.0x more |
| Claude Sonnet 5 Anthropic → | $2.00 | $10.00 | 3.4x more |
| Gemini 3.1 Pro Google → | $2.00 | $12.00 | 4.0x more |
| Gemini 3.5 Flash Google → | $1.50 | $9.00 | 3.0x more |
| Claude Haiku 4.5 Anthropic → | $1.00 | $5.00 | 1.7x more |
| DeepSeek V4 Pro DeepSeek → | $0.435 | $0.870 | 3.1x less |
| DeepSeek V4 Flash DeepSeek → | $0.140 | $0.280 | 9.7x less |
“Same job” = a 1,000-token prompt with a 3,000-token reply, priced on both models. Cheaper is not always better. It is, however, always cheaper.
NOW FIND OUT HOW MUCH OF THAT IS WASTE
Gemini 3 Flash's price is fixed. Your token count isn't. Paste a real prompt into the Burn Calculator and see what it costs on Gemini 3 Flash — and how much a leaner version would save at 1,000 runs. Or pull the whole board from the ticker.