One-Line Summary

Google cuts the cost of agentic AI with Gemini 3.7 Flash while pushing everyday automation into Workspace and Spark, as OpenAI deepens enterprise routes via IBM and Anthropic pursues compute efficiency with a potential Decart deal.

Big Tech

Google launches Gemini 3.7 Flash with focus on coding and agents

Google releases Gemini 3.7 Flash, a lower-cost “workhorse” model designed to write code and run multi-step agents, with an introductory API price of $0.75 per million input tokens and $3.75 per million output tokens through Dec 31, 2026; the release lands just three weeks after 3.6 Flash. Google attributes the upgrade to developer feedback and algorithmic improvements. ¹

On Google’s published tests, 3.7 Flash raises FrontierCode 1.1 Main to 43.6% (from 34.4%) and DeepSWE v1.1 to 65.3% (from 49.0%). It also improves web app generation on WebDev Arena (Elo 1588 vs 1538) and handles knowledge-dense tasks better on GDP.PDF (34.0% vs 22.0%) and AutomationBench (30.4% vs 17.0%). ¹

Google says the model adapts more effectively to roadblocks, clarifies intent, and follows instructions with greater fidelity—aiming to cut retries and manual oversight in coding and business agents. The 50% price cut applies through the end of 2026, with pricing returning to $1.50 (input) and $7.50 (output) per million tokens on Jan 1, 2027. ²

Gemini Spark, Google’s personal agent for Google AI Pro and Ultra subscribers in over 160 countries, moves to 3.7 Flash starting Aug 13, 2026, improving tool use across Google Workspace; enterprises can access the model via Gemini Enterprise Agent Platform and the Gemini Enterprise app. Google also ships updated safeguards for chemical, biological, radiological, and nuclear risks and cyber-offense misuse. ¹

IBM partners with OpenAI to expand enterprise deployments

IBM says it is partnering with OpenAI to co-market offerings and build industry-specific solutions across financial services, government, telecom, and retail; IBM will train and certify tens of thousands of consultants on OpenAI’s tech and integrate models including GPT-5.6, Codex, and ChatGPT Work into IBM Consulting Advantage and its Autonomous Security service. ³

OpenAI appoints Dali Rajic as Chief Revenue Officer to build a global revenue operating system, noting its products reach more than 1 billion weekly active users and over 2 million businesses; Denise Dresser will depart after a transition period. ⁴

OpenAI also outlines how enterprise use is shifting from assistance to execution: as of June, Codex accounted for 64% of combined Codex and ChatGPT output tokens among enterprise customers, and “frontier firms” generated 8.3× as many output tokens per active user as typical firms. ⁵

Industry & Biz

Anthropic explores $6B acquisition of Decart to boost efficiency

Anthropic is in talks to acquire Israeli startup Decart AI for about $6 billion, in what would be its largest known acquisition; the deal is not finalized and could still fall through, according to reporting. ⁶

Reports indicate Decart builds software that helps chips run AI more efficiently and develops video-simulation technology; sources say a deal could bring those chip-efficiency and simulation capabilities into Anthropic’s inference work. ⁶

Calcalist adds that Decart raised $300 million in May led by Radical Ventures at roughly a $4 billion valuation with Nvidia participating, and that its tools include Lucy for real-time video editing and Oasis for generating simulated environments. ⁷

New Tools

Google Sheets canvas turns spreadsheets into mini‑apps with Gemini

Google introduces Sheets canvas, a Gemini-powered feature that converts spreadsheet tabs into interactive “mini-apps” from a simple prompt—no formulas or code required—with real-time sync to the underlying sheet and full collaboration. ⁸

Examples include study trackers, dashboards, fantasy rosters, and seating charts. The feature is available globally in English to Google AI Pro and Ultra subscribers and is rolling out to Workspace Business and Enterprise Standard and Plus plans, and to Google AI Pro for Education add-ons. ⁸

Community Pulse

Hacker News (946↑) — Users praise Gemini Flash for reliability and speed as a daily assistant and are curious whether 3.7 improves further. ⁹

"I have had such good, consistent success with Gemini 3.6 Flash (usually on medium setting) that I forget about it. It is just this reliable, effective assistant that I use for managing and organizing information, brainstorming, and of course coding. I honestly don't know what I would want more/better than what I was already getting... so I'm curious if I will see any improvements with 3.7." — Hacker News ⁹

What This Means for You

If you run coding agents, web-generation, or document-processing tasks, the temporary 50% price cut for Gemini 3.7 Flash changes unit economics. Don’t just compare price per million tokens—test cost per successfully completed task; Google claims fewer retries and higher first-pass quality, which can lower total spend. ²

For non-developers, agents get easier to use in familiar apps. Spark now runs on 3.7 Flash for Google AI Pro and Ultra subscribers and can consolidate files, draft emails, and update status docs across Workspace—useful for PMs, marketers, and ops teams. ¹

Teams can also build lightweight internal tools without engineering help. Sheets canvas turns any spreadsheet into a working dashboard or tracker in minutes, staying in sync with your data and shareable as a normal tab. ⁸

Enterprise buyers will see more packaged routes to deployment. IBM’s partnership gives OpenAI broader implementation muscle, while Anthropic’s interest in Decart highlights a race to improve compute efficiency—two forces that can make agent projects cheaper and faster to stand up. ^{3 6}

Action Items

1. Trial Gemini 3.7 Flash in AI Studio: Run two of your top workflows (code fix, PDF summary) and record time, retries, and tokens to compare cost per completed task against your current model.
2. Turn a Sheet into a working app: Open Google Sheets, use the Ask Gemini side panel’s “Create canvas” to build a live dashboard or tracker synced to your existing data.
3. Set up one Spark workflow: If you have Google AI Pro or Ultra, have Spark consolidate files, draft an email, and update a status document for a current project.
4. Book a 60‑minute scoping call with IBM Consulting: If you work with IBM, request details on its new OpenAI integration and pick one workflow for a two‑week proof‑of‑concept with ChatGPT Work or Codex.