1 week ago
Google's Gemini 3.7 Flash Gains Speed Before Price Doubles
Google released a new AI model called Gemini 3.7 Flash.
It arrived only three weeks after the previous version.
Tests showed that it performed much better at coding and completing tasks.
It also produced answers very quickly in independent testing.
Google is offering it at low introductory prices until the end of 2026.
On January 1, 2027, those prices will double.
Companies using the model may then need to raise their prices, accept lower profits, or switch models.
Other companies are competing with faster or open-weight models, but Gemini is broadly available now.
Google released Gemini 3.7 Flash on August 13, three weeks after Gemini 3.6 Flash.
The model improved on FrontierCode 1.1 Main, DeepSWE v1.1, and AutomationBench benchmarks.
Artificial Analysis ranked it first among 186 measured models for output speed, at 340.1 tokens per second.
Introductory pricing is $0.75 per million input tokens and $3.75 per million output tokens through December 31, 2026.
Prices double to $1.50 and $7.50 per million tokens on January 1, 2027, potentially raising costs for companies that adopt it.
- Who
- Google released Gemini 3.7 Flash, while OpenAI, Cerebras, and Moonshot are cited as competitors.
- What
- Gemini 3.7 Flash delivered large benchmark and speed improvements but has introductory pricing that doubles in 2027.
- Where
- The article discusses the international AI-model market; no specific release location is given.
- When
- It was released on August 13; current prices run through December 31, 2026, and double on January 1, 2027.
- Why
- Google appears to be encouraging adoption with low introductory prices while disclosing a scheduled increase.
Introductory Pricing Opportunity
Adoption and Cost Risk
Value for AI developers
Introductory Pricing Opportunity
Gemini 3.7 Flash is presented as fast, inexpensive, and generally available, giving developers a low-cost way to build products during the introductory period.
Adoption and Cost Risk
The low rate is temporary, so companies basing their products on it may face a significant cost increase after December 31, 2026.
Pricing disclosure
Introductory Pricing Opportunity
Google published the future price increase in advance, providing more notice than is usual in the AI industry.
Adoption and Cost Risk
Advance notice does not remove the risk that companies will have to raise prices, absorb lower margins, or switch models after integrations are already in place.
Competitive position
Introductory Pricing Opportunity
Gemini 3.7 Flash combines strong benchmark gains with broad availability and high measured speed.
Adoption and Cost Risk
OpenAI's Ultrafast mode for GPT-5.6 Sol reportedly reaches up to 750 output tokens per second, while Moonshot's open-weight Kimi K3 can be self-hosted, although their availability or pricing differs.
Key facts
- Release date
- August 13
- Predecessor
- Gemini 3.6 Flash, released three weeks earlier
- FrontierCode 1.1 Main
- 43.6%, compared with 34.4% for Gemini 3.6 Flash
- DeepSWE v1.1
- 65.3%, compared with 49.0% for Gemini 3.6 Flash
- AutomationBench
- 30.4%, compared with 17% for Gemini 3.6 Flash
- Current price
- $0.75 per million input tokens and $3.75 per million output tokens
- Price from January 1, 2027
- $1.50 per million input tokens and $7.50 per million output tokens
- Artificial Analysis speed ranking
- First among 186 measured models, at 340.1 tokens per second
Quotes
Company that sets Gemini pricing
“Those are introductory rates. They run until December 31, 2026. On January 1, 2027, they become $1.50 and $7.50 — exactly double.”
wionews.com
Sources
Related news
Google Launches Gemini 3.7 Flash; Flagship Pro Remains Delayed
DeepSeek launches V4 Pro at prices 14 times V4 Flash
OpenAI Cuts GPT-5.6 Sol Developer Prices Amid Competition
OpenAI Cuts Frontier GPT-5.6 Sol Developer Prices Over 20%
AI Model Releases Outpace Independent Testing, Raising Evaluation Concerns
China Challenges US Leadership in Escalating Global AI Contest
Trump-backed crypto firm backs AI venture offering restricted Chinese models
OpenAI's Ultrafast Mode Makes GPT-5.6 Sol 14 Times Faster
Alibaba plans revenue share for users of Qwen open-source model
