Study guide
5.2 Claude Model Family Selection
Claude Certified Developer - Foundations CCDV-F
Track your progress as you read
Finished sections get a check mark.
1 Chapter 1: Core Messages API & Streaming
1.1 Messages API Fundamentals & Request Lifecycle 1.2 Streaming & Server-Sent Events (SSE) 1.3 Thinking: Adaptive Mode, Effort & Reasoning Budgets
2 Chapter 2: Optimization & Asynchronous Batches
2.1 Prompt Caching Mechanics & Breakpoints 2.2 Message Batches API & Asynchronous Workloads
3 Chapter 3: Structured Outputs, Documents & Files API
3.1 Structured Outputs & JSON Extraction 3.2 Files API, Document Blocks & Citations
4 Chapter 4: Application Design, Resilience & Delivery Life Cycle
4.1 API Resilience, Error Handling & Retries 4.2 Production Backend Architecture & State Management 4.3 Requirements Analysis & Claude Application Design 4.4 The Claude Application Life Cycle & Configuration Management
5 Chapter 5: LLM Fundamentals, Model Selection & Versioning
5.1 LLM & Technical Fundamentals: Tokens, Context Windows & Sampling 5.2 Claude Model Family Selection 5.3 Model Versioning, Snapshot Pinning & Migration 5.4 Latency & Throughput Optimization
6 Chapter 6: Cost Engineering & Context Management
6.1 Production Cost Optimization Strategies 6.2 Context Window Management & Compaction
7 Chapter 7: Foundations of Workflows & Deterministic Patterns
7.1 Agent vs. Workflow Architectural Foundations 7.2 Workflow Patterns: Prompt Chaining & Routing 7.3 Workflow Patterns: Parallelization & Voting
8 Chapter 8: Autonomous Agents & Multi-Step Systems
8.1 Workflow Patterns: Orchestrator-Workers & Evaluator-Optimizer 8.2 Agent Loop Implementation & State Tracking 8.3 Agent Frameworks & Human-in-the-Loop (HITL)
9 Chapter 9: Prompt and Context Engineering
9.1 System Prompt Design & Role Framing 9.2 XML Tags & Structural Prompt Delimiters 9.3 Few-Shot Prompting & In-Context Learning 9.4 RAG Context Injection & Document Formatting
10 Chapter 10: Tool Calling & Model Context Protocol (MCP)
10.1 Tool Schema Design & Parameter Validation 10.2 Tool Execution, Multi-Tool Calls & Error Handling 10.3 MCP Architecture & Core Primitives 10.4 MCP Transports & Client/Server Integration 10.5 Agentic Customization: Built-in Tools, Custom Tools, Skills & MCP
11 Chapter 11: Security, Safety & Governance
11.1 Direct & Indirect Prompt Injection Defenses 11.2 Tool Permissions, Sandboxing & Least Privilege 11.3 Data Privacy, Redaction & Audit Logging 11.4 Identity, Secrets & API Key Management
12 Chapter 12: Claude Code Developer Tooling
12.1 Claude Code Architecture & Project Configuration 12.2 Permissions, Modes & Hooks
13 Chapter 13: Evaluation, Testing & Production Observability
13.1 Evaluation Design & Golden Datasets 13.2 Production Observability, Tracing & Debugging
5.2 Claude Model Family Selection
Key Takeaways Explain This Topic AI
The current Claude API lineup is Claude Haiku 5.5 (from $0.10/$0.50 per MTok), Claude Sonnet 5.5 ($2/$10), Claude Opus 5.5 ($4/$20), and Claude Fable 5.1 ($10/$50); Claude Opus 5, Claude Sonnet 5, and Claude Haiku 4.5 are legacy models that remain available, while Claude 3 Opus, Claude 3.5/3.7 Sonnet, and Claude Haiku 3.5 are retired.
Every current model carries a 1M-token context window and 128K max output; the 200K window and 64K output cap belong to older models such as the legacy Claude Haiku 4.5. Long prompts are standard-priced except on Claude Haiku 5.5, where prompts over 100,000 tokens cost more.
Adaptive thinking is on by default on every current model, and depth is steered with output_config.effort (low, medium, high, xhigh, max); the default is medium on Claude Opus 5.5 and Claude Haiku 5.5 and high on Claude Sonnet 5.5 and Claude Fable 5.1. Among older models still available, Claude Haiku 4.5 and the deprecated Claude Sonnet 4.5 lack the effort parameter.
Model selection balances four vectors: task cognitive complexity, latency SLA, unit economics per transaction, and accuracy criticality - and within one model, adjusting effort is a cheaper first lever than changing tier.
Routing sends each request to a tier-appropriate handler up front, while cascading runs the cheap tier first and escalates on a measurable signal; cascading only pays off when the cheap tier deflects a large majority of traffic.
Last updated: October 2026
Claude Model Family Selection (Opus, Sonnet, Haiku)
Exam Blueprint Focus: Model Selection and Trade-offs is its own weighted sub-skill inside the 16.8% Model Selection and Optimisation domain. CCDV-F expects you to match a stated business requirement to the right Claude tier using capability, latency, and unit economics — and to know that Anthropic ships one current lineup at a time , so a design that names a retired model is wrong regardless of how well it reasons.
Architectural Philosophy of the Claude Tiered Hierarchy
Anthropic ships the Claude family in named tiers rather than one monolithic model. The tier names — Haiku , Sonnet , Opus , and the research-grade Fable line — are stable across generations; the generation number attached to each name is what moves. Reading a tier name alone tells you the intelligence-per-dollar bracket. Reading the full model ID tells you exactly which weights you are calling.
Every current model supports the same platform primitives: the Messages API , vision (image input), tool calling, prompt caching, the Message Batches API, and citations. They diverge on three axes:
Reasoning depth — how much multi-step deduction the model sustains without drifting.
Latency and throughput — time-to-first-token (TTFT) and sustained tokens per second.
Unit economics — dollars per million input and output tokens.
The Current Lineup
The table below is the current lineup as published in Anthropic's models overview on 26 October 2026. Prices are per million tokens (MTok) on the first-party Claude API.
- Specification Claude Haiku 5.5 Claude Sonnet 5.5 Claude Opus 5.5 Claude Fable 5.1
- Claude API model ID claude-haiku-5-5 claude-sonnet-5-5 claude-opus-5-5 claude-fable-5-1
Positioning High-volume, latency-sensitive tasks such as classification, extraction, and routing Best combination of speed and intelligence Long-running agentic coding and knowledge work Demanding reasoning and long-horizon agentic work
- Input price From $0.10 / MTok $2 / MTok $4 / MTok $10 / MTok
- Output price From $0.50 / MTok $10 / MTok $20 / MTok $50 / MTok
- Cache read 0.1x base input 0.05x base input 0.05x base input 0.025x base input
- Context window 1M tokens 1M tokens 1M tokens 1M tokens
- Max output (sync) 128K tokens 128K tokens 128K tokens 128K tokens
- Thinking Adaptive, on by default Adaptive, on by default Adaptive, always on Adaptive, always on
- Default effort medium high medium high
- Comparative latency Fastest Fast Moderate Slower
Claude Haiku 5.5 prices apply to prompts up to 100,000 tokens; longer prompts cost $0.50 / $2.50 per MTok. Legacy models remain available (Claude Opus 5 at $5/$25, Claude Sonnet 5 at $2/$10, Claude Haiku 4.5 at $1/$5 with a 200K window and 64K output cap), but a design described as "current" should name the lineup above.
Two facts decide most exam questions:
Context window. Every current model, Haiku 5.5 included, carries the full 1M-token window. A scenario that says "we must hold a 400,000-token codebase in one request" is now decided by capability, latency, and cost rather than by window size; only legacy models such as Haiku 4.5 are capped at 200K.
effort . On every current model, output_config.effort is the primary intelligence/cost dial. Its default differs by model ( medium on Opus 5.5 and Haiku 5.5, high on Sonnet 5.5 and Fable 5.1), so set it explicitly rather than relying on the default.
1. Claude Haiku 5.5 — Volume, Latency, and Guardrails
Haiku 5.5 is the throughput tier. Use it where the unit of work is small, well-specified, and repeated thousands of times per hour:
Classification and routing. Intent detection, ticket triage, language detection, and the front-door router in a cascading architecture.
Extraction at volume. Pulling three fields out of a receipt; tagging metadata; reformatting.
Safety and guardrail passes. A cheap second model that scores a response for policy violations before it reaches the user (see the dual-LLM pattern in the security chapter).
Reranking. Scoring 50 retrieved chunks for relevance before the expensive model reads the top 8.
Trade-offs: Haiku 5.5 trades reasoning depth for speed and price, and prompts over 100,000 tokens move to its higher price band. Use low effort for chat, short tool tasks, and simple high-volume requests; Anthropic notes that in long agent prompts the model is more likely to skip a search or stop early at low . It is not the right host for a long autonomous agent loop that must plan across dozens of files.
2. Claude Sonnet 5.5 — Speed and Intelligence
Sonnet 5.5 is the best combination of speed and intelligence in the lineup. At $2/$10 per MTok it costs half of Opus 5.5 while carrying the same 1M-token window, the same 128K max output, adaptive thinking, and the full low – max effort range (default high ). It fits scenarios that say "high quality, cost matters, and we serve real users":
Customer-facing chat and support agents with a latency budget.
Retrieval-augmented generation over large document sets.
Tool-using agents of moderate depth.
Code generation and review inside a CI budget.
3. Claude Opus 5.5 — Long-Running Agentic and Knowledge Work
Opus 5.5 is Anthropic's suggested starting point when you are unsure which model to use, and the tier for work where a wrong answer is expensive and the task is genuinely hard: long multi-file refactors, deep architectural planning, orchestrator roles that decompose work for cheaper workers, and analysis where the reasoning chain is the deliverable. At $4/$20 per MTok it is 2x Sonnet 5.5.
Opus 5.5 has two exam-worthy behavioural rules: thinking cannot be disabled at any effort level ( thinking: {"type": "disabled"} returns a 400 error ), and its effort default is medium , one level lower than Claude Opus 5's high . Forced tool_choice ( any or tool ) also returns a 400 error on Opus 5.5.
4. Claude Fable 5.1 — Frontier Reasoning
Fable 5.1 sits above Opus 5.5 for demanding reasoning and long-horizon agentic work; thinking is always on and cannot be turned off. At $10/$50 per MTok it is the most expensive tier, and its cache reads are priced at 0.025× base input. Reach for it when evals on Opus 5.5 at higher effort still fall short — not as a general-purpose default.
- Production Decision Framework: Four Vectors
- Vector 1 — Task cognitive complexity
Ask how many dependent inference steps the task needs. Single-step extraction or classification → Haiku 5.5. Multi-step synthesis with a known shape → Sonnet 5.5. Open-ended planning, long tool chains, or genuinely novel analysis → Opus 5.5, then Fable 5.1 if evals justify it.
Vector 2 — Latency SLA
Map the SLA to the tier, then to effort . A tight latency budget on a keystroke-latency feature points at Haiku 5.5 at low effort. A conversational assistant points at Sonnet 5.5 — Anthropic suggests starting chat and other latency-sensitive work on Sonnet 5.5 at medium or low effort before you consider changing model. Stream the response (SSE) so the user sees TTFT, not total generation time.
Vector 3 — Unit economics
Compute cost per transaction, not cost per token. A 20,000-token cached system prompt read on every call costs $0.002 per call on Sonnet 5.5 (20,000 × $2 × 0.05 / 1,000,000), $0.004 on Opus 5.5 (20,000 × $4 × 0.05 / 1,000,000), and $0.0002 on Haiku 5.5 (20,000 × $0.10 × 0.1 / 1,000,000). Multiply by request volume before choosing. Remember the two structural discounts that apply on every tier: cache reads at 0.1× base input or less (0.05× on Opus 5.5 and Sonnet 5.5, 0.025× on Fable 5.1) and the Batch API at 50% off both input and output .
Vector 4 — Accuracy criticality
Where an error creates legal, financial, or clinical exposure, buy the more capable tier and add a verification pass. The cost of one bad contract summary dwarfs the delta between Sonnet 5.5 and Opus 5.5 on a few thousand documents.
- Multi-Model Architectures: Routing and Cascading
- Pattern 1 — Intent routing
A cheap classifier reads the request and dispatches it to a tier-appropriate handler. Claude Haiku 5.5 at low effort with structured outputs (or a classification tool; Haiku 5.5 still accepts a forced tool_choice ) is a natural router: it returns a category quickly for a fraction of a cent, and the expensive model only sees the requests that need it.
Pattern 2 — Model cascading (fallback escalation)
Run the cheap tier first, then escalate on a measurable signal — a self-reported confidence score below threshold, a failed schema validation, or a refusal. Cascading pays off when the cheap tier resolves a large majority of traffic; if it only resolves half, you pay twice on the rest and the blended cost can exceed calling Sonnet 5.5 directly. Caches are scoped per model, so a cascade also forfeits cache reuse across its tiers. Measure the deflection rate before committing to the pattern.
Pattern 3 — Orchestrator/worker split
Opus 5.5 plans and decomposes; Haiku 5.5 or Sonnet 5.5 workers execute the leaf tasks in parallel; Opus 5.5 synthesises. You pay frontier prices only for the planning and synthesis tokens.
Common Traps & Antipatterns
Naming a retired model. Claude 3 Opus, Claude 3.5 Sonnet, Claude 3.7 Sonnet, and Claude Haiku 3.5 are retired on the first-party Claude API. An answer choice built on them is wrong even if the reasoning is sound.
Calling a legacy model "current." Claude Opus 5, Claude Sonnet 5, and Claude Haiku 4.5 are still available but are legacy; the current lineup is Fable 5.1, Opus 5.5, Sonnet 5.5, and Haiku 5.5.
Assuming the fast tier has a small window. Haiku 5.5 has the full 1M window; the 200K cap belongs to legacy models such as Haiku 4.5.
Assuming one effort default. Omitting effort runs medium on Opus 5.5 and Haiku 5.5 but high on Sonnet 5.5 and Fable 5.1.
Reaching for a bigger model when the real fix is effort or a better prompt. Adjusting effort within one model is cheaper and faster to test than a tier change.
Ignoring the Haiku 5.5 long-prompt band. Long context is standard-priced on the other current models, but Haiku 5.5 prompts over 100,000 tokens cost more.
Defaulting to Opus 5.5 "for quality." Sonnet 5.5 at the right effort clears many production bars at half the price; run the eval before choosing.