Google Reportedly Shelves Gemini 3.5 Pro

Bloomberg reported on July 16 that Gemini 3.5 Pro was months behind schedule because Google was working to improve its capabilities, particularly coding, citing people familiar with the matter. The report said internal stakeholders were concerned about competitive pressure as Anthropic and OpenAI released models that exceeded Gemini's capabilities in some areas.

CNBC separately reported that Alphabet had introduced Gemini 3.5 Pro at its May Google I/O conference, where it said the model was in internal use and was expected to reach a broader rollout the following month. After Bloomberg's report, an Alphabet spokesperson told CNBC that Google was "shipping quickly across a wide range of models while keeping them highly cost-effective for customers." The spokesperson added that the company was testing 3.5 Pro, an upgraded Flash model, and other models with partners.

That public statement is important context for the newer RSS claim: as of the July reporting, Google had described Gemini 3.5 Pro as being in partner testing, rather than discontinued. No retrieved primary Google statement or independently retrieved report confirms the reported shelving decision, its rationale, or a changed availability status.

On July 21, Google released Gemini 3.6 Flash, Gemini 3.5 Flash-Lite, and the security-focused Gemini 3.5 Flash Cyber, according to CNET and Mashable. Google characterized 3.6 Flash as a general-purpose "workhorse" model with improvements in coding, knowledge work, and multimodal tasks, CNET reported.

CNET reported that Google claimed up to 17% lower token usage for 3.6 Flash than its predecessor, alongside lower per-token costs. Mashable reported list pricing of $1.50 per million input tokens and $7.50 per million output tokens for 3.6 Flash. Google also said 3.5 Flash-Lite could produce 350 output tokens per second, according to CNET, while Flash Cyber was designed for vulnerability-finding and remediation workflows with the CodeMender infrastructure agent.

At the time of that release, Google stated that Gemini 3.5 Pro was "currently testing with partners" and would reach general availability "as soon as it's ready," Mashable reported. The same Google update said training runs for a future Gemini 4 model had begun.

For teams choosing models, a reported cancellation and a delayed partner-testing program represent materially different procurement and migration risks. Public API availability, pricing, benchmark documentation, and product release notes are more reliable operational signals than uncorroborated reports about internal model roadmaps.

More broadly, the July releases illustrate a common model-provider pattern: lower-latency and lower-cost variants can arrive while a higher-capability flagship remains under evaluation. Practitioners evaluating Gemini for coding or agentic systems can compare the published Flash pricing and performance claims, but should not treat Gemini 3.5 Pro's reported cancellation as confirmed without a Google announcement or additional independent reporting.