GPT-4o Interview Questions

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.

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 GPT API question on structured outputs reliability vs JSON mode.

**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.

**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.

**Structure** Eval templates + data JSONL + scoring functions; CI integration via API.

**Model selection** Benchmark retrieval on your corpus — smaller models cheaper if recall OK.

**Fit** Nightly evals, backfill embeddings, document tagging — not interactive chat.

**Model behavior** o-series and GPT-4.1 emphasize extended internal reasoning before answer — improved reliability on complex tasks, not always needed for chat.

**Common failures** Model calls nonexistent tool; args wrong type; parallel tool_calls partial failure; streaming partial JSON; user content tricks model into exfil tool args.

**Concept** Responses treat interaction as structured items (messages, tool calls, outputs) with native support for reasoning models and built-in tools — reducing glue code.

**Assistants API** Managed threads, code interpreter, file search, function tools with run lifecycle (queued/in_progress). Less boilerplate for CRUD-style agent apps. Trade-offs: vendor lock-in, run polling complexity, harder fine-grained graph control.

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

**Pipeline** Pre-filter user input → model → post-filter output before display; async moderation for streaming with revoke.

Multi-agent orchestration is a senior/staff-level system design question gaining traction at Google DeepMind and Microsoft. Learn the supervisor pattern, shared state management, and how to avoid the 'too many agents' anti-pattern that sinks most candidate answers.

Structured JSON output from LLMs is a production necessity at Stripe, Vercel, and every AI-native startup. Interviewers test your reliability engineering — schema enforcement, retry logic, streaming edge cases — not just 'use JSON mode.' Master the full production playbook.

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

**Eval dimensions** SWE-bench-style tasks on private repo sample; CI fix rate; hallucinated APIs.