Sundar Pichai admits Google trails in AI coding

2026-07-23 · Source: Dataconomy · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Software Development & Engineering · Depth: Intermediate, quick

Summary

Alphabet CEO Sundar Pichai stated during the company's Q2 2026 earnings call that Google needs to improve its coding and agentic coding capabilities, emphasizing a larger Gemini 4 base model is crucial for competitiveness. These comments followed the introduction of Gemini 3.6 Flash and acknowledged delays for the unreleased Gemini 3.5 Pro due to coding issues. Internally, these delays led to the departure of senior AI researchers Noam Shazeer and John Jumper. Google recently launched Gemini 3.6 Flash, which produces 17% fewer output tokens than 3.5 Flash, enhancing cost-effectiveness and coding abilities. On the DeepSWE coding benchmark, 3.6 Flash scored 49%, up from 37% with 3.5 Flash. Gemini 3.5 Pro is still in partner testing, while Gemini 4 remains in pretraining, representing Google's most ambitious effort yet.

Key takeaway

For AI Product Managers evaluating large language model adoption, Google's acknowledged lag in AI coding and Gemini 3.5 Pro delays suggest caution. You should assess competitor offerings, especially those with strong agentic coding capabilities and proven developer-facing products. Consider integrating Gemini 3.6 Flash for immediate efficiency gains, but factor in Google's ongoing challenges when planning long-term strategic dependencies on their unreleased flagship models.

Key insights

Google acknowledges trailing in AI coding, stressing larger base models and improved agentic capabilities are vital for future competitiveness.

Principles

• Larger base models drive breakthroughs.
• Developer-facing products yield crucial usage data.
• Coding performance impacts model release timelines.

In practice

• Integrate Gemini 3.6 Flash for efficiency.
• Prioritize agentic coding development.
• Monitor DeepSWE benchmark for coding progress.

Topics

• AI Coding
• Agentic Coding
• Gemini Models
• Large Language Models
• DeepSWE Benchmark
• Google AI Strategy

Best for: CTO, VP of Engineering/Data, AI Architect, Director of AI/ML, AI Product Manager, Tech Journalist

Related on AIssential

• What's going on with Gemini? — Google's Gemini strategy prioritizes internal token consumption and hardware integration over direct competition in external frontier model intelligence.
• OpenAI's GPT 5.5 is wild... — Advanced AI coding capabilities are driving a recursive improvement cycle across major AI labs.
• Sergey Brin commits DeepMind to a Claude catch-up — Improving AI's coding ability is seen as the shortest path to achieving self-improving AI systems.
• The AI Awards 2025 - Best LLM? Biggest Moment in AI? Best Agentic Coder? — Personalized AI awards for 2025 highlight top models, platforms, and companies across key categories.
• TAI #193: Gemini 3.1 Pro Takes the Benchmarks Crown, but Can it Catch Up in the Tools Race? — Raw model intelligence does not always equate to practical utility in real-world applications.
• Google CEO Pichai says Gemini's next leap depends on building "much larger base models" — Future frontier AI capabilities hinge on developing "much larger base models" and a full-stack approach to AI solutions.

See Counsel's argued verdicts on the open AI decisions leaders are weighing →

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Dataconomy.