Grok V9-Medium 1.5T Coding Model, Grok 4.3 on Bedrock with 1M Context, Agent Dashboard and Databricks Integration

Grok V9-Medium: xAI releases 1.5-trillion-parameter coding-focused model (June 16)

• xAI released Grok V9-Medium on June 16, a 1.5-trillion-parameter model specifically optimized for coding tasks.
• making it the largest publicly available Grok model to date and positioning it as a direct competitor to frontier coding models from Anthropic and OpenAI.

Editor’s read: 1.5T parameters for coding is a significant bet - benchmark it against your current stack before committing to a model switch.

Grok 4.3 now generally available on Amazon Bedrock: 1M token context, configurable reasoning (June 17)

• Grok 4.3 went GA on Amazon Bedrock on June 17 with a one-million-token context window and configurable reasoning depth.
• allowing developers to tune the model's reasoning budget per request; available via standard Bedrock APIs with no special access required.

Editor’s read: Configurable reasoning on Bedrock means you pay only for the reasoning depth your task actually needs - set it low for classification tasks, high for complex code generation.

Grok Imagine Video 1.5: sharper realism, better physics simulation, faster generation (June 17)

• xAI launched Grok Imagine Video 1.5 on June 17, announcing improvements in photorealistic output quality, physics simulation accuracy, and generation speed.
• relevant for developers using Grok's multimodal capabilities in product demos or prototyping workflows.

Grok Build gains Agent Dashboard and Priority Processing tier (June 17)

• The Grok Build platform added an Agent Dashboard for monitoring active agent runs and a Priority Processing tier that provides lower-latency inference during high-load periods, as reported in the week of June 15-17 summary.

xAI Grok lands on Databricks at 2026 Data + AI Summit (June 18)

• xAI announced a Grok integration with Databricks at the 2026 Data + AI Summit on June 18, enabling enterprise data teams to invoke Grok models directly within Databricks Unity Catalog workflows and notebooks.

Grok for Microsoft 365 and expanded platform integrations announced (week of June 16-21)

• The Grok Build June 2026 summary reports a Grok for Microsoft 365 integration alongside the Bedrock and Databricks expansions, extending xAI's enterprise reach into the Microsoft productivity stack alongside its cloud infrastructure play.

levelsio on what to build as an indie developer in 2026 (June 15)

• On June 15, Pieter Levels shared his framework for indie developer focus in 2026, arguing niche specificity and distribution matter more than ever as model quality converges and generic AI products commoditize.

Editor’s read: Good prompt for Grok Build users: which of your product ideas uses Grok's unique strengths (speed, large context, configurable reasoning) rather than just substituting it for another model?

Rauchg: Vercel launches Fluid compute on microVMs with no function runtime ceiling (June 15)

• Vercel Fluid compute, announced June 15 by Guillermo Rauch, removes the traditional serverless function runtime ceiling via microVMs.
• enabling the longer-running agentic tasks that Grok Build workflows increasingly require.

SpaceX acquires Cursor for $60B in stock - Musk gains a second AI coding platform alongside Grok (June 16)

• On June 16, SpaceX announced the acquisition of Cursor, the AI IDE, for approximately $60 billion in stock, days after SpaceX's IPO; the deal gives Elon Musk control of both xAI/Grok and Cursor, creating an unusual competitive dynamic where a single owner now controls two competing AI developer tool ecosystems.

Editor’s read: Watch whether post-acquisition Cursor is steered to prefer Grok models - if so, the Grok/Cursor stack becomes an integrated play worth evaluating.

Fable 5 still suspended globally on Day 9 after US export control order (June 21)

• Anthropic's Fable 5 and Mythos 5 remain offline as of June 21, suspended by a US government export control order issued June 12-13; restoration talks between Anthropic and Trump officials have been complicated by an NSA breach claim.

GLM-5.2: Z.ai's 753B open-weights model beats GPT-5.5 on coding and tops Artificial Analysis leaderboard (June 17)

• Simon Willison reported June 17 that Z.ai's GLM-5.2.
• a 753-billion-parameter open-weights model.
• topped the Artificial Analysis coding leaderboard and outperformed GPT-5.5, introducing a serious self-hostable alternative to proprietary frontier models.

Microsoft AutoJack: malicious web page can hijack an AI coding agent and execute arbitrary code on the host (June 18)

• Microsoft disclosed AutoJack on June 18, a prompt-injection technique where a single malicious web page can take over a running AI coding agent and achieve remote code execution on the developer machine; applicable to any Grok Build agent with web tool or browser access.

Editor’s read: If your Grok agent fetches external URLs or browses the web, this disclosure applies directly - sandbox your runs before your next demo.