During Alphabet’s Q2 2026 earnings call, CEO Sundar Pichai said Google needs to improve its coding and agent capabilities, and that a larger Gemini 4 model will be needed to stay competitive in the next frontier.
These comments come a day after Google launched Gemini 3.6 Flash and announced that Gemini 4 is currently in development. Meanwhile, the unreleased Gemini 3.5 Pro is still delayed due to coding issues.
What is said by Photosi
When asked if Gemini could stay on the frontier, Photosi expressed confidence that Google is still working well in many areas but noted that coding and encoding needs more improvement.
He pointed out the recent launch of 3.6 Flash as a good step, while explaining the future plans, explaining that reaching the next breakthrough depends on the big base models. Google is currently training Gemini 4, which Photosi says Google will need to compete at that level.
This is consistent with his earlier views. In May, Photosi said on the Hard Fork podcast that Google was “a bit behind” in agent coding due to the lack of a developer-oriented product that generates valuable usage data collected by competitors.
Flagship Still Not Out
Google’s Gemini 3.5 Pro hasn’t shipped yet. The company announced the 3.5 series at I/O in May, introduced the 3.5 Flash the same day and said the 3.5 Pro will follow next month.
Google now says that 3.5 Pro is “currently in testing with our partners” and will be widely available “as soon as it’s ready.” Bloomberg reported this month that code enforcement contributed to the delay, as did the SEJ. A late June update to the model’s training data, intended to improve coding, did not meet expectations.
The pressure was felt elsewhere. In June, two of Google’s top AI researchers, Gemini co-lead Noam Shazeer and John Jumper of AlphaFold, left for OpenAI and Anthropic, amid concerns within DeepMind about Google’s position on AI coding tools.
Posted by Google
Google recently released Gemini 3.6 Flash, an update to the more reliable Flash tier than the flagship model. The company claims that 3.6 Flash generates 17% fewer output tokens compared to 3.5 Flash and is more cost-effective, with improved coding capabilities.
| Model | Current Situation | Intended Role | Time |
|---|---|---|---|
| Gemini 3.5 Flash | Available | A general-purpose Flash model | Released at Google I/O |
| Gemini 3.6 Flash | Available | Workhorse updated with coding and performance improvements | Released on July 21 |
| Gemini 3.5 Flash-Lite | Available | A fast, low-cost model for high-volume cargo | Released on July 21 |
| Gemini 3.5 Pro | Testing with colleagues | Flagship model in the 3.5 series | No release date has been confirmed |
| Gemini 4 | In training | The next generation base model | No release date has been confirmed |
In DeepSWE, the coding benchmark Google cited, it scored 49%, up from 37% with 3.5 Flash. Google describes 3.6 Flash as the new standard workhorse, while 3.5 Flash is still available. They also introduced the affordable 3.5 Flash-Lite category, now integrated into Google Search.
Why This Matters
More important is the timeline than how Google compares to others. They’ve already introduced Flash models in AI Mode and Search, but 3.5 Pro has slipped, and Gemini 4 doesn’t have an official date yet.
Looking Forward
The most important thing now is whether the 3.5 Pro finally goes away and if Google sticks to the monthly release schedule mentioned by Photosi. Successfully hitting these goals will show how Google can turn its plans into real, shipped products.
On the other hand, Gemini 4 is a long-term goal. Google calls this pre-training process its biggest ambition so far, but they haven’t set a release date for the model yet.
Featured Image: Thrive Studios/Shutterstock ID
