GPT vs Gemini

OpenAI's model lineup changes fast. GPT-4 vs GPT-4o is a model selection question that tests whether you understand latency, cost, multimodal capabilities, and when reasoning depth matters. Critical for any role touching OpenAI APIs in production.

Google's Gemini 1.5 Pro long-context window opens use cases impossible with standard LLMs — whole-codebase analysis, multi-hour video, massive document review. Interviewers test whether you understand real limitations behind the 1M token marketing number.

Codex CLI represents OpenAI's push into agentic coding outside the editor — similar category to Claude Code and other CLI agents. The core loop is: understand task → explore repo → edit → run commands → iterate.

Cost optimization for coding agents is context engineering + runtime policy.

Medium Gemini/Vertex question on context caching vs RAG trade-offs.

Medium GPT API question on structured outputs reliability vs JSON mode.

**Session model** WebSocket/WebRTC streaming multimodal input/output.

**Flow** Multimodal intake → diagnose → suggest fix → verify resolved.

**Vertex settings** Safety filters on HATE, HARASSMENT, etc. with threshold BLOCK_ONLY_HIGH vs stricter.

**Hybrid** Traditional OCR for text layer + Gemini for semantic extraction.

**Model choice** Flash for simple; Pro for hard multimodal only when needed.

**Ingestion** Transcode standards; chunk very long media if API limits apply.

**API** Tool declarations with JSON schema; multi-turn functionResponse loop.

**Infrastructure** Managed endpoints vs batch prediction; GPU quotas.

**Mechanism** API retrieves search results, model grounds answer with sources.

**Gemini** Long video understanding, Google ecosystem integration.

**Use cases** Full codebase questions, hours-long media, multi-doc merger analysis.

**Storage** Vector + structured facts table keyed by user_id; version memories on conflict.

**Proxy layer** Backend holds keys; assigns budgets per customer/plan.

**Azure benefits** VNet, Azure AD auth, Microsoft compliance certifications, existing EA billing.

**Contractual** Enterprise DPA, ZDR endpoints, prohibition on training customer data.

**Session** Persistent realtime connection with session.update for instructions/tools; handle reconnect.

**Use cases** Where zero-shot vision fails consistently on proprietary visual patterns.

**Decision** Try prompt + tools + RAG first; fine-tune if style/reliability gap persists and data exists.