AI News Daily 2026-09-22
- xAI announced Grok 4.7, its new flagship model, raising parameter count from Grok 4.6's 1.5 trillion to 2.1 trillion and targeting stronger performance on coding, agentic workflows, and specialist-knowledge tasks.
- Despite the upgrade, xAI held pricing steady at $2 per million input tokens and $6 per million output tokens — the same rate as Grok 4.6 — making the new model an immediate drop-in upgrade for cost-conscious teams using Cursor, Grok Build, and the xAI API.
- Microsoft's Q2 2026 global AI diffusion report found worldwide AI usage among the working-age population at 18.8%, up roughly one point quarter-on-quarter, while the gap between Global North (28.8%) and Global South (16.2%) adoption rates widened further.
01 xAI unveils Grok 4.7, a new flagship model built for coding agents
Published: 2026-09-21 09:00
Facts
xAI announced a new flagship model, Grok 4.7. The model's parameter count grows from Grok 4.6's 1.5 trillion to 2.1 trillion, and xAI says the increase — combined with longer reinforcement-learning runs and a heavier emphasis on harder tasks — is aimed at improving performance on coding, agentic workflows, and specialist-knowledge tasks. Pricing is unchanged from Grok 4.6, at $2 per million input tokens and $6 per million output tokens. The model is available through Cursor, Grok Build, and the xAI API.
Background
Grok 4.7 succeeds Grok 4.6 in xAI's flagship line, continuing a pattern of iterative model upgrades focused specifically on coding-agent and agentic-workflow performance rather than general-purpose capability alone.
Implications
Because xAI kept pricing flat while expanding model scale and reported capability, the practical effect for teams already budgeting around Grok 4.6 rates is a capability upgrade at no additional cost. For organizations evaluating AI coding assistants, this is worth watching as a case of a vendor improving the cost-performance ratio of an existing price tier rather than introducing a new, more expensive one.
02 Microsoft's latest AI diffusion report shows a widening North-South adoption gap
Published: 2026-09-21
Facts
Microsoft published the latest edition of its global AI diffusion report, covering the second quarter of 2026. It found that AI usage among the working-age population averaged 18.8% worldwide, up roughly one percentage point from the first quarter. The United Arab Emirates (73.3%) and Singapore (64.3%) remained the top-ranked markets. At the same time, the report states that the gap between Global North adoption (28.8%) and Global South adoption (16.2%) widened further compared with the prior quarter.
Background
Microsoft has been tracking AI usage rates across countries on a recurring basis as part of its ongoing "state of global AI diffusion" reporting series, positioning UAE and Singapore consistently among the highest-adoption markets in earlier editions referenced by this report.
Implications
The report's finding that regional adoption gaps are widening, rather than narrowing, as overall global usage rises is a relevant data point for multinational organizations reassessing AI rollout strategy and workforce planning on a region-by-region basis.
Microsoft On the Issues — The Continued State of Global AI Diffusion in 2026
03 Editor's note
Today's two confirmed items point in complementary directions: one is a vendor-side capability release (Grok 4.7) that lowers the effective cost of frontier-level coding-agent performance, and the other is a macro-level usage report (Microsoft's diffusion data) showing that access to and adoption of AI capability remains highly uneven across regions even as overall usage climbs. Read together, they suggest that while the technical frontier keeps advancing at flat or falling cost, the benefits are not yet spreading evenly worldwide.
Coverage for this cycle was comparatively light: no new pricing or terms-of-service changes were confirmed from major providers during the collection window, and other reports still awaiting independent confirmation were not included here. Readers should expect a fuller slate in the next edition as pending items are verified.