Model Picker
Last reviewed: June 2026
Model names and pricing change frequently. Verify against provider docs before production.
Full guides: Cost, Latency, and Tokens · Choosing a Tool · Tokens and Context
Production API Tiers (2026)
| Tier | Examples | Use when |
|---|---|---|
| Fast / cheap | Claude Haiku, GPT-4o-mini, Gemini Flash | Classification, routing, simple Q&A, high volume |
| Balanced | Claude Sonnet, GPT-4o | Most chat, coding help, tool calling |
| Premium / reasoning | Claude Opus, o-series, GPT-4.1 | Hard bugs, architecture, low-volume critical tasks |
Default: balanced tier. Escalate to premium only when balanced fails twice or stakes are high.
Task → Model (Production)
| Task | Model tier | Notes |
|---|---|---|
| Intent classification / routing | Fast | One-line label output |
| Customer support FAQ | Balanced | Add RAG for grounding |
| Code generation in product | Balanced | Cap maxTokens; verify output |
| Complex analysis / legal / medical | Premium + human review | Never auto-act |
| Embeddings | Embedding model | Not chat models |
| Batch summarization (offline) | Fast or Batch API | Cheaper than interactive |
Dev Tooling (Cursor, Copilot, Claude Code)
| Task | Mode / model | Notes |
|---|---|---|
| Tab completions | Default / fast | Accept line by line |
| Explain code | Ask + balanced | Attach @Files |
| Multi-file feature | Plan → Agent + balanced/premium | Approve plan first |
| Stuck bug (2+ failures) | Debug or premium | Paste full error output |
| Boilerplate scaffold | Balanced or fast | Verify imports after |
| Security-sensitive code | Human-led | AI drafts only; strict review |
| Docs / comments | Fast or balanced | Verify facts |
When to Switch Models Mid-Session
| Signal | Switch to |
|---|---|
| Repeated wrong API hallucinations | Premium + @Docs or official docs pasted |
| Agent over-refactors | Stop agent; Plan mode with file boundaries |
| Simple rename across files | Fast model or inline Copilot |
| Architecture decision | Ask mode premium; you decide and implement |
Cost Quick Math
Monthly cost ≈ (requests × avg_input_tokens × input_price)
+ (requests × avg_output_tokens × output_price)
| Tactic | Saves |
|---|---|
| Truncate chat history | Input tokens |
maxTokens cap | Output tokens |
| Prompt caching (Anthropic/OpenAI) | Repeated system prompt |
| Fast model for routing | Routes 80% of traffic cheaply |
| Rate limits per user | Abuse / runaway loops |
Provider Cheat Sheet
| Provider | Strong at | Watch out for |
|---|---|---|
| Anthropic | Long context, tool use, coding | Model ID strings change |
| OpenAI | Ecosystem, vision, embeddings | Pricing tiers shift |
| Google Gemini | Multimodal, GCP integration | Region availability |
| Local (Ollama) | Privacy, offline | Quality vs cloud frontier |