Gemini 3.5 Pro missed its expected June rollout. An anonymous-source account says a late-June data change fell short of Google's coding goals, but Google has confirmed only partner testing—not the reported cause, a new date, or public results and pricing.
Google has missed its expected June rollout for Gemini 3.5 Pro. That is established. The reported explanation—a coding shortfall that persisted after a late change to the data used to train Gemini—has not been independently established: Google has confirmed testing, but not the cause, internal targets or a new launch date.
That distinction matters because most coverage traces to one underlying account. Repetition across many summaries does not independently corroborate the employees who described the model work.
Google introduced the Gemini 3.5 family on May 19, released Gemini 3.5 Flash and said Pro was already in internal use. Its announcement said the company looked forward to rolling Pro out “next month.” A review of the announcement and API release record found no Pro release after June passed and no replacement month.
The claim that Pro is “months behind schedule” refers to an internal timetable, not the public one. The central account originated with Bloomberg. A summary of that reporting said 10 current and former employees described frustration among engineers, researchers and managers, with some worried that OpenAI and Anthropic were moving ahead.
The same account said Google changed the data used to train Gemini in late June to improve coding, but the results fell short of expectations. A detailed relay of the account said performance in other domains remained unclear. The evidence therefore supports continued model work and a reported coding problem; it does not establish that Google discarded a base model, that coding is the only weakness or that Pro failed broadly.
A Google spokesperson said in a statement responding to the report that the company was testing 3.5 Pro, an upgraded Flash model and other models with partners, and was engaging with the US government on model testing and broader frameworks. Google also said it was shipping a wide range of models while keeping them cost-effective. It did not confirm the reported coding result or give a date.
Gemini 3.1 Pro, released in February, consequently remains the available Pro model. Gemini 3.5 Flash is the only publicly released model in the 3.5 line.
The delay is consequential because coding products are already sold through subscriptions, APIs and enterprise platforms. Google CEO Sundar Pichai had also acknowledged that the company was “a bit behind” the frontier in agentic coding, according to the retained account. That is narrower than saying Gemini trails in every coding task, but it puts the reported Pro problem in context.
Google has commercial exposure before Pro ships. Its $100-a-month AI Ultra subscription targets developers, technical leads and other advanced users, with five times the Gemini app and Antigravity usage limit of the Pro plan. Google says its compute-based limits treat complex coding prompts as more demanding than simple text. The subscription bundles models, tools, storage and other benefits; it is not a disclosed price for Gemini 3.5 Pro.
OpenAI made GPT-5.6 Sol, Terra and Luna available across ChatGPT, Codex and its API on July 9. Its published API prices per million tokens are $5 input and $30 output for Sol, $2.50 and $15 for Terra, and $1 and $6 for Luna. Those figures establish availability and vendor pricing, not superior value in every workload. They are also not directly comparable with Google's bundled consumer subscription.
Meta has released Muse Spark 1.1, while other available coding alternatives include open-weight models. But the rival-release story needs its own qualification. OpenAI's GPT-5.6 public launch followed a delay tied to US government requests, and Anthropic temporarily disabled its most advanced models after a US export-control order before restoring access with safeguards, according to the same industry account. Model availability is shaped by regulation and safety controls as well as engineering speed.
Employees described multiple stakeholder layers around model releases and integrations spanning Search, Maps and YouTube, which they said can slow launches, in a summary of the internal account. Other reporting described separate coding efforts in DeepMind's AI Studio, Google Cloud's Vertex AI and Android Studio, an attempt to unify internal coding tools and limits on the compute available to some engineers for internal AI use.
None of those facts shows that organizational structure or compute access caused the reported coding miss. They describe possible release friction around a company with an unusually broad product surface.
Google's own adoption figure adds another limit. As of April, the company said 75% of new code was AI-generated and then approved by engineers, up from 50% the previous fall. Former employees also described resistance from engineers who believed important code should remain human-written. The percentage measures assisted output that passed review, not autonomous software production, and it does not identify which models produced the code.
Alphabet shares traded down after the delay report, although snapshots differed: one put the decline near 3%, while another recorded 4%. The move shows market sensitivity to the timetable, but it does not measure subscriber losses, foregone revenue or the value of the unreleased model.
Gemini 3.5 Flash already serves customers. It is also the default model for AI Mode, and Google's materials did not say Pro would replace it, the release analysis noted. The Pro delay therefore does not by itself change the answers Google Search serves today.
Nor does it show that every Gemini product is stalled. Flash remains available, and Google says Pro testing continues with partners. What the delay withholds is the evidence needed to judge the flagship: customers cannot test its reported coding improvements, inspect its latency and reliability or compare its operating price with shipped alternatives.
A release date would settle only when Gemini 3.5 Pro becomes available. Resolving the central question—whether Google closed the reported coding gap—requires public model access, clearly scoped coding evaluations, API and subscription pricing, rate limits, latency and results from real developer workflows.
Google would also need to distinguish the effect of model quality from release review, safety requirements, compute constraints and organizational handoffs. Until those facts are public, “testing with partners” establishes activity, while claims about the model's competitiveness remain untestable.
OpenAI has put conversations and Projects back in its redesigned ChatGPT desktop app and enabled cloud Work threads to move across devices, correcting the launch's biggest usability failures without merging local Work or Codex histories.
Meta reportedly plans to put departing AWS compute executive Dave Brown to work on its data-center buildout, adding hyperscale operating experience while leaving any customer-facing cloud business conditional and undefined.
China has paired a five-year AI training offer for developing countries with cooperation centers, a weather-warning rollout and a new 29-country organization. The package gives Beijing a platform for influence, but no budget, selection rules or delivery timetable has been published.
RoboTTT extends a robot policy’s visuomotor context to 8,000 timesteps and raises its average rubric score from 42% to 79%, but the evidence comes from three author-run assembly tasks and the longest task was completed in only two of ten trials.
Zhipu reportedly reached $1 billion in annual recurring revenue in July, roughly four times a March estimate, but the unconfirmed run rate is not annual sales and still sits far ahead of recognized cloud revenue while margins remain thin.
Nebius has arranged its first senior secured facility against an operating GPU deployment and one customer’s cash flows. The deal adds project-level debt to its expansion toolkit, but an unnamed customer and a larger rival financing limit what it proves about the rest of Nebius’s backlog.
Cursor says automatically running a repository-root git.exe on Windows does not meet its criteria for patching, while the researcher calls it an untrusted-search-path defect and separate research shows the same weakness across several competing AI coding tools.
Huawei publicly displayed a 16-cabinet Atlas 950 configuration rated at 1 EFLOPS in FP8, providing tangible evidence of its system-scale AI strategy while leaving price, power use and sustained workload performance undisclosed ahead of the full system's planned fourth-quarter release.
CIA Director John Ratcliffe said US intelligence is consistent with an estimate that Russian recruits last 20 to 30 minutes on Ukraine’s battlefield, but the public trail leads to an unsourced claim about assault troops and does not establish a representative average.
UK testing places leading open-weight models four to seven months behind selected closed-model cyber results, yet longer attack chains, U.S. benchmarks and mixed cost comparisons show why that interval is a warning signal rather than a universal capability clock.
TSMC reached the top of its second-quarter revenue guidance and raised its 2026 outlook, but a one-off investment gain boosted profit growth while the company committed more capital to 2-nanometer production, advanced packaging and an undated Arizona expansion.
Moonshot AI's Kimi K3 added to a global technology selloff with near-frontier performance, but unreleased weights, mixed cost comparisons and a recommended 64-accelerator deployment leave its effect on chip demand unresolved.
Moonshot AI has made Kimi K3 available through its apps and API, pairing a 2.8-trillion-parameter architecture with early frontier-level results, but the model's open-weight claim cannot be tested until its weights and technical report arrive.
Two binding EU decisions require Google to give rival AI services comparable access to 11 Android features and offer eligible search competitors a restricted, anonymized dataset, but phased deadlines, certification, pricing and privacy safeguards leave the competitive effect unproven.
Databricks has signed a term sheet for a Coatue-led financing at a $188 billion valuation, while unidentified sources put the round at $3 billion. The proposed capital would deepen its push into AI governance, data agents and operational databases, but the transaction remains open and the company supplied no new operating figures.
Netflix says generative AI workflows were used on roughly 300 titles in 2026, but its only quantified example tied to that disclosure covers 17 minutes and does not establish how much finished material, spending or labor changed across the slate.
Google is bringing Gemini Omni editing and a reusable face-and-voice avatar to Vids, but the sharper distinction is account-level identity across Vids and Gemini rather than a new category of AI video; Vids already offered Veo generation and customizable avatars, while specialist rivals already sell digital presenters.
Intel plans to adopt Gemini Enterprise across engineering, supply-chain and corporate work while extending chip-simulation capacity into Google Cloud, but the companies supplied no rollout schedule, contract value or performance baseline.
The Navy has approved an immediate department-wide framework for turning data into operational effects, but the public rollout leaves budgets, owners, deadlines, performance measures and assurance rules to a promised implementation roadmap.
Runta has raised a $20 million seed round led by Andreessen Horowitz for an execution layer that governs agents at the operating-system and network levels, entering a market where cloud runtimes and sandbox providers already offer overlapping controls.