Google has introduced Gemini 3.7 Flash within weeks of the previous version’s launch, delivering significant improvements in coding, automation, and document processing while slashing prices for users.
Google has moved quickly to replace Gemini 3.6 Flash with a newer version inside Gemini Spark, and the upgrade is not a minor one. The company says Gemini 3.7 Flash is stronger at coding, automation and reading PDF documents, while also arriving on a lower promotional price than the model it supersedes.
According to La Razón, Gemini 3.6 Flash launched on 21 July and was overtaken in Spark on 13 August for AI Pro and Ultra subscribers, leaving only 23 days between the two versions in that product. Google has kept 3.6 Flash available through its API, so the older model has not been removed entirely.
The clearest gains appear in tasks that require agents to operate tools rather than merely answer questions. In Terminal-bench 3.0, which tests terminal-based work, Gemini 3.6 Flash scored 5.4 per cent, while 3.7 Flash reached 14.9 per cent. In AutomationBench, a benchmark for multi-step workflow handling, the score rose from 17 per cent to 30.4 per cent.
Programming performance has also improved materially. On DeepSWE v1.1, the model advanced from 48.6 per cent to 65.3 per cent, while FrontierCode 1.1 moved from 34.4 per cent to 43.6 per cent. For developers, that kind of change is practical: better first-pass code should reduce retries and cut the number of prompts needed to reach usable output.
Document handling is another area where Google appears to have concentrated effort. On GDP.pdf, a test for understanding complex PDF files, the score increased from 22 per cent to 34 per cent. That suggests a meaningful lift for users asking Gemini to work through long reports, manuals or other dense documents, although La Razón notes that Gemini 3.7 Flash does not beat its predecessor on every benchmark, with CharXiv one example where it trails slightly.
Pricing is the other major change. Google is charging 75 cents per million input tokens and $3.75 per million output tokens for Gemini 3.7 Flash, which the company says is half the original launch price of Gemini 3.6 Flash. The same promotional rates now apply to 3.6 Flash as well, but only until 31 December 2026. From 1 January, both models are due to rise to $1.50 per million input tokens and $7.50 per million output tokens.
Disclaimer: This content is intended for informational purposes only. Readers are advised to exercise their own judgement, conduct due diligence, or consult a qualified expert before acting on any information provided.
