Gemini 3.7 Flash

This Ofox blog post dated 2026-09-03 compares Gemini 3.8 Flash against DeepSeek V4 Flash 0731 but provides material context on Gemini 3.7 Flash as the immediate predecessor. It states Google shipped Gemini 3.8 Flash on 2 September 2026 at the same per-token price as 3.7 Flash, and that 3.7 Flash itself shipped on 13 Au For developers budgeting on 3.7 Flash via Ofox, the article specifies the introductory rate of $0.75 per million input tokens and $3.75 per million output, the same as 3.7 Flash, expiring on 31 December 2026. It warns that pricing doubles to $1.50/$7.50 per million on 1 January 2027, so all Ofox Gemini 3.7 Flash cost n

Google's September 1, 2026 monthly recap explicitly names the Gemini 3.7 Flash launch among its biggest August 2026 AI announcements, alongside Gemini 3.5 Transcribe and the Pixel 11 series. The post frames August as focused on making AI "more practical through new hardware... and cost-efficient models like Gemini 3.7 The recap adds useful context by situating Gemini 3.7 Flash within Google's wider August 2026 product push, including the rollout of Gemini in Chrome on Android, hands-free voice tools across Workspace and Gemini Live, studio-quality video and music generation, and open-source weather/climate AI models. It reiterates t

Emergent.sh describes Gemini 3.7 Flash as Google's newest workhorse model for coding and agents, released on August 13, 2026, just three weeks after Gemini 3.6 Flash. It is an algorithmic refinement of 3.6 Flash rather than a new base model, with its biggest gains in coding, agent workflows, and document comprehension. The supplied excerpt lists a 1M-token context window, text/image/audio/video/PDF inputs, text-only output, and introductory pricing of $0.75 per 1M input tokens and $3.75 per 1M output tokens through December 31, 2026, doubling to $1.50 and $7.50 on January 1, 2027. The article frames Flash releases as accelerating fas

Aiwiz.uk's August 22, 2026 piece frames Gemini 3.7 Flash — launched August 13, 2026 — primarily as a coding, web development, and multi-step agent upgrade, but addresses its applicability to everyday business workflows. The article argues the model is a credible option for analyzing marketing data, preparing reports, c Google's documentation lists the model as generally available, but the article cautions that the best current public automation result remains below one task in three under strict scoring. It recommends a human-in-the-loop pattern — using 3.7 Flash to prepare, classify, compare, and check work, while people retain cont

This NYU Shanghai RITS analysis characterizes Gemini 3.7 Flash as a three-week point release on the same architecture and training data as 3.6 Flash, with the model card explicitly describing it as algorithmic improvements to the core reasoning foundation rather than a new pretraining run. Specifications carried over i The article documents coding and agentic gains versus 3.6 Flash: FrontierCode 1.1 Main moves from 34.4% to 43.6% (reportedly above Claude Sonnet 5 at 42.7% and GPT-5.6 Terra at 41.3%), DeepSWE v1.1 from 48.6% to 65.3%, Code Arena Elo from 1538 to 1588 (top of Google's comparison set), AutomationBench from 17.0% to 30.4

Miraflow published a benchmark-focused breakdown on August 16, 2026, of Google shipped Gemini 3.7 Flash on August 13, 2026 at half the price of 3.6 Flash, citing Google's official launch post. The article framed the gains as post-training and reinforcement learning on agentic trajectories rather than scale or a longer The piece highlighted the DeepSWE long-horizon coding benchmark jump from 49.0% to 65.3% and noted that the model introduces a new "thinking level" parameter that replaces the older "thinking budget" control surface. It promised a hands-on API walkthrough with runnable code and discussed common production mistakes team

Google released Gemini 3.7 Flash on August 13, 2026, describing it as "our most intelligent workhorse model yet for coding and agents," with a stable model ID of gemini-3.7-flash (no preview suffix). The model supports a 1,048,576-token context window and a 65,536-token max output, accepting text, image, video, audio, Google-reported benchmark deltas versus Gemini 3.6 Flash highlight coding and agentic gains: FrontierCode 1.1 rises from 34.4% to 43.6%, DeepSWE v1.1 from 49% to 65.3%, Terminal-bench 2.1 from 78.0% to 85.8%, AutomationBench from 17.0% to 30.4%, WebDev Arena from 1538 to 1588 Elo, and GDM-MRCR v2 at 128k from 91.8% to

Gemini 3.7 Flash shipped as a generally available model under the identifier gemini-3.7-flash, framed by Google as "our most intelligent workhorse model yet for coding and agents." It carries a 1M token context window with a 64k output ceiling, accepts text, image, video, audio, and PDF input, but produces text-only ou The same review documents breaking API changes that accompany the 3.7 release: developers migrating from Gemini 3.5 Flash, Gemini 3 Flash Preview, or Gemini 3.1 Pro are told to strip temperature, top_p, top_k, and candidate_count, replace the numeric thinking budget with a thinking-level string, remove prefilled model

AIToolly documents Google DeepMind's official announcement of Gemini 3.7 Flash on August 13, 2026, via the DeepMind Blog, positioning it as the newest member of the Flash category optimized for speed and efficiency. The release continues the established Gemini lineage of scalable AI solutions for varied applications. The piece frames the move from version 3.6 to 3.7 as a steady progression in Google's model development strategy, reflecting the industry-wide trend of frequent incremental updates. It reiterates that the Flash series traditionally balances high-speed processing with sophisticated reasoning for developers and enterpris

Ars Technica reports that Google announced Gemini 3.7 Flash on August 13, 2026, rolling it out as the replacement for Gemini 3.6 Flash just three weeks after that predecessor's release. Senior Director Tulsee Doshi is quoted framing the model as a "workhorse" iteration built from core optimizations and developer feedba The piece also catalogs the benchmark deltas Google is using to justify the rapid release cadence. On FrontierCode 1.1 Main, Gemini 3.7 Flash rises from 34.4% to 43.6%, and DeepSWE v1.1 climbs from 49% to 65.3%, per the cited figures. Document handling improves on GDP.pdf (22% to 34%), WebDev Arena Elo rises to 1,588 f