Observed Signal · Aug 4, 2025 · Technical Release · Source: OnlineMarketing.de · Impact: 4/5 · Sentiment: Positive
Google's Deep Think: More Thinking Time for Better Answers
Google's Gemini 2.5 gains a Deep Think feature that extends the AI's thinking time by running multiple reasoning paths in parallel and evaluating their quality before delivering a final answer. Deep Think relies on a longer inference phase and reinforcement learning to encourage the model to use these extra steps, aiming for answers that are more thorough, creative, and strategically valuable across tasks from iterative design to mathematics. The capability is available now to Google AI Ultra subscribers in the Gemini app as part of Gemini 2.5 Pro. Early tests suggest Deep Think excels on tasks requiring multiple iterations, such as design refinement and complex development projects, with demonstrations including a Pagodenvergleich showing varying detail levels across Gemini 2.5 Flash, Pro, and Deep Think. Google claims improvements in research support, coding, and mathematical reasoning, supported by benchmarks like LiveCodeBench V6 and Humanity’s Last Exam, while cautioning about potential inaccuracies and the need for human oversight.
Technical release from major platform (Google) enabling deeper thinking in Gemini 2.5; potential industry impact.
Track Google Signals & Market Shifts in Real-Time
Polaris7 autonomous intelligence agents track regulatory filings, primary sources, executive changes, and deal flow 24/7. Create your free Explorer workspace to monitor these entities.
Key Takeaways & Evidence Grounding
- Deep Think extends Gemini 2.5's thinking time by running parallel thought processes.
- It evaluates multiple solution paths and synthesizes the best into a final answer.
- Available now to Google AI Ultra subscribers in the Gemini App under Gemini 2.5 Pro.
- Based on longer inference time and reinforcement learning to encourage extra thinking steps.
- Benchmarks and demonstrations claim improved performance in logic, coding, and mathematics, including a Pagodenvergleich showing three scenes of increasing detail.
Connected Companies & Entities
2 Entities mappedOntology Mapping & Concepts
Related Market Signals & Shifts
Recent verified developments and strategic activity across this market segment.
Gemini 3.1 Pro: Advanced reasoning for complex tasks
Google has rolled out Gemini 3.1 Pro, the latest in its Gemini AI family, designed to handle tasks where simple answers aren’t enough. Google claims 3.1 Pro delivers roughly double the reasoning performance of the November 2025–released Gemini 3 Pro, unlocking capabilities for complex data synthesis, visual explanations, and creative visuals in UX design and marketing workflows. The update follows Gemini 3 Deep Think, further emphasizing agentic workflows. Rollout is in Preview via the Gemini App for AI Ultra and Pro subscribers, with NotebookLM support, and API access through AI Studio, Antigravity, Vertex AI, Gemini Enterprise, Gemini CLI, and Android Studio. It’s also available in other environments such as the Perplexity AI Answering Machine. The article highlights use cases like generating data tables, creating animations, and even analyzing literary topics (e.g., Wuthering Heights) to inform websites or portfolios, illustrating Google’s push toward more capable, reasoning-focused AI tools.
Google Launches Gemini 3.8 Live and Extended Thinking Audio Models
Google has introduced two new AI audio models, Gemini 3.8 Live and Gemini 3.8 Live Extended Thinking, designed for near-zero-latency, real-time conversations and complex multi-step tasks. The models support parallel processing of speech, visual inputs, and function calls, ensuring uninterrupted interactions. They are available via the Gemini Live API and Google AI Studio, with pricing at $0.005 per minute for audio input and $0.018 for output. Gemini 3.8 Live is integrated into Search Live for camera-based interaction, while Extended Thinking is incorporated into Gemini Live and Workspace (Docs, Gmail, Keep) for Google AI subscribers. All generated audio is watermarked with SynthID for transparency. Independent tests show strong performance, with Extended Thinking scoring 82.6 on the Speech-to-Speech Index. Additionally, Google highlighted Gemini 3.5 Transcribe, a streaming speech-to-text model with a 4.0% word error rate.
Google Expands Gemini Notebook with Visible Thinking Steps
Google has renamed and expanded NotebookLM into Gemini Notebook, a more integrated research workspace that connects the Gemini app, Google Search and Google Workspace. New user-facing features include expandable "thinking steps" for eligible Google AI Ultra and Pro subscribers, in-notebook code execution, and continued citation-backed outputs (Audio Overviews, Mind Maps). The rollout is phased (initial access for Google AI Ultra and Workspace AI Ultra Access, then Pro on the web) with mobile expansion planned. Google says uploaded sources remain private and are not used to train models; the company frames thinking steps as experimental and an inspection aid rather than proof of correctness.
Track Real-Time Market Signals & Shifts
Set up custom watchlists to receive automated, evidence-grounded executive digests whenever material signals or shifts occur across your tracked landscape.