Gemini
productGemini
Type: Product (Large Language Model Family)
Gemini is Google DeepMind's family of large language models. In this knowledge base, it appears primarily as the engine powering AlphaEvolve's evolutionary code generation.
The AlphaEvolve system uses a Flash/Pro ensemble strategy: Gemini Flash maximizes the breadth of ideas explored (generating many candidate solutions quickly), while Gemini Pro provides critical depth with insightful suggestions (evaluating and refining the most promising candidates).
The 2026 upgrade to Gemini 2.5 Pro enabled semantic evolution — the model can now reason about code logic and control flows rather than just adjusting parameter values. This qualitative leap allows AlphaEvolve to discover fundamentally different algorithmic approaches.
Gemini 3.7 Flash and the Dated Price (Aug 2026)
This page has until now tracked Gemini only as the engine inside AlphaEvolve. The 2026-08-13 GA of Gemini 3.7 Flash adds a second reason to track it: Google is the clearest current case of a provider publishing a future price as a dated commitment.
Read from Google's own pricing page on 2026-08-18, the paid standard tier for Gemini 3.7 Flash is quoted with the date inline:
| Item, per Mtok | Through 2026-12-31 | From 2027-01-01 |
|---|---|---|
| Input | $0.75 | $1.50 |
| Output (incl. thinking tokens) | $3.75 | $7.50 |
| Context caching | $0.075 | $0.15 |
| Cache storage (per Mtok-hour) | $0.50 | $1.00 |
Every element doubles, on one date, published at launch.
The part that is easy to miss: Google applied the identical dated rate to the incumbent. Gemini 3.6 Flash was recorded in the Token Price Index at a flat $1.50 / $7.50 on 2026-08-03. On 2026-08-18 it reads $0.75 / $3.75 through December 31, 2026. $1.50 / $7.50 starting January 1, 2027 — the same words as 3.7 Flash. So Google did not merely launch a new model below the rung; it cut the incumbent 50% to meet the new model and attached a single shared expiry to the whole live Flash generation.
Prior-generation Gemini 3.5 Flash is untouched at a flat $1.50 / $9.00 with no date attached, which is the control that makes the reading safe: the promotion is scoped to the current generation, not applied line-wide, and it is not a page-wide formatting artifact.
Evidence: strong (primary — Google's own Gemini API pricing page, read 2026-08-18, quoted verbatim). One gap: the date Google cut 3.6 Flash is not established. The Index's prior observation is 2026-08-03 and the next is 2026-08-18, so the change is bounded to that window, not dated. The 2026-08-13 GA date for 3.7 Flash is secondary (wire coverage); the pricing page carries no release date.
Key Contributions
- Flash/Pro ensemble for evolutionary search: Breadth (Flash) + depth (Pro) strategy in AlphaEvolve (AlphaEvolve)
- Gemini 2.5 Pro semantic evolution: Logic/control flow rewriting, not just hyperparameter tuning (AlphaEvolve)
- Self-improving loop: AlphaEvolve optimizes Gemini's own training kernels (23% speedup, 1% training time reduction) (AlphaEvolve)
Mentioned In
- Evolutionary Algorithm Discovery — Powers the evolutionary search
- LLM Tool Use — Code generation as tool use
Related Entities
- Google DeepMind — Developer
- AlphaEvolve — Primary application documented in this KB
Changelog
- 2026-08-18 — Added the Gemini 3.7 Flash GA and its dated pricing. Recorded the finding this page did not previously hold: Google cut the incumbent Gemini 3.6 Flash 50% to the new model's introductory rate and put one shared 2027-01-01 doubling across both, while leaving prior-generation 3.5 Flash on a flat undated rate. All rates read verbatim from Google's own pricing page on 2026-08-18. Widens this entity from "the engine inside AlphaEvolve" to also being the KB's cleanest example of a published future price. No AlphaEvolve content changed.