What will Gemini 3.5 Pro's standard API INPUT price be, in USD per 1,000,000 tokens at the ≤200k-context tier, at public release?

Gemini 3.5 Pro, announced at Google I/O 2026, is expected to reach General Availability in June or July 2026 3 sources. The fundamental tension in pricing this model is the conflict between Google's established price-performance leadership and the upward pressure caused by its own new Flash tier.

Currently, Gemini 3.1 Pro sits at $2.00 per 1M input tokens (≤200k context) 44 sources. However, Gemini 3.5 Flash recently launched at $1.50 per 1M input tokens—a 3x jump from the previous Flash generation [the-decoder, llm-stats, de54c9]. Holding 3.5 Pro flat at $2.00 would severely compress the tiers, making a modest price increase highly probable to maintain product separation.

Conversely, Google heavily markets Gemini as significantly cheaper than comparable frontier models blog.google. GPT-5.4 currently charges $12 for input, and Claude Opus sits around $15. To maintain its undercutting strategy, Google is incentivized to keep its marketing-visible input prices relatively low. This is particularly relevant given the integration of 'Deep Think' reasoning; because reasoning effort is billed as flat-rate output tokens, Google can capture significant revenue on the backend without drastically raising the sticker price of input tokens. A WaveSpeed assessment similarly floats a modest ~$2.50 input price for a premium reasoning tier wavespeed.ai.

The probability distribution reflects these opposing forces:

• Lower Bound (p10 at $2.00): Anchored to the status quo. Google has historically held Pro input prices flat across generations ai.google.dev, and extreme price stickiness remains plausible if they prioritize aggressive market share acquisition.
• Center (p25–p50 at ~$2.75–$4.33): The base case assumes a moderate generational uplift. A ~$4.33 median creates the necessary separation from the $1.50 Flash tier while still undercutting GPT-5.4 by more than 60%, perfectly aligning with Google's historical positioning.
• Right Tail (p75–p90 up to $13.50): Accommodates a specific secondary-source rumor floating a ~$15 input / $60 output price aiweekly.co. While matching Opus pricing would mean abandoning Google's explicit price-undercutting strategy, it cannot be completely discounted given the compute requirements of a 2M-context window and Deep Think integration. The upper percentiles also account for the possibility of Google maintaining historical Pro-to-Flash pricing ratios (~4x), which would imply an input price of roughly $6.00 to $7.00.