5.3 Model selection and tradeoffs
The current line-up
| Model | ID | Context | Input $/MTok | Output $/MTok |
|---|---|---|---|---|
| Claude Fable 5 | claude-fable-5 | 1M | $10.00 | $50.00 |
| Claude Opus 5 | claude-opus-5 | 1M | $5.00 | $25.00 |
| Claude Opus 4.8 | claude-opus-4-8 | 1M | $5.00 | $25.00 |
| Claude Sonnet 5 | claude-sonnet-5 | 1M | $2.00 | $10.00 |
| Claude Haiku 4.5 | claude-haiku-4-5 | 200K | $1.00 | $5.00 |
Use the exact ID strings — no date suffixes. Prices are first-party API rates; Bedrock and Vertex are partner-operated with separate pricing.
Choosing a tier
| Tier | Fits |
|---|---|
| Opus (and Fable, above it) | Hard reasoning, long-horizon agentic work, complex coding, ambiguous judgment, the orchestrating agent in a hierarchy |
| Sonnet | The broad middle — production workloads that need strong quality at lower cost, high-volume extraction and analysis |
| Haiku | High-volume, latency-sensitive, well-specified work: classification, routing, simple extraction, cheap subagents. Note the 200K window |
Three tradeoff axes, and they do not move together:
| Axis | Lever, cheapest first |
|---|---|
| Cost | Prompt caching → prune input → shorter output → batch → lower effort → smaller model |
| Latency | Streaming (time-to-first-token) → shorter output → lower effort → smaller model → fast mode on Opus 5/4.8 |
| Quality | Better prompt and tool descriptions → higher effort → larger model → decomposition into passes |
Before building a multi-model cascade, measure the more capable model at lower effort on the same tasks. Lower effort on a newer model often matches or beats a prior generation at high effort, and one model means one cache namespace — caches are model-scoped, so a cascade forfeits cache reuse across its models. Judge cost per completed task, not cost per request: a cheaper call that needs three retries is not cheaper.
Capability differences that change code
| Capability | Note |
|---|---|
| Adaptive thinking | {"type": "adaptive"} on all current models. On Claude Opus 5 thinking is on by default when thinking is omitted; on Opus 4.8/4.7 omitting it means no thinking. On Fable 5 it is always on and cannot be configured |
| Effort levels | low–max on current models; xhigh arrived with Opus 4.7 |
| Sampling parameters | Removed on Opus 5 / 4.8 / 4.7, Sonnet 5, Fable 5 — 400 if sent |
budget_tokens | Removed on those models; deprecated on the 4.6 family |
| Assistant prefill | Removed on 4.6 and later — 400 if sent. Use structured outputs instead |
| Fast mode | Claude Opus 5 / Opus 4.8 only, first-party only, premium pricing |
| Context window | 1M on current models; 200K on Haiku 4.5, which also caps inline PDFs at 100 pages |
| Max output | Up to 128K on current models; streaming required for large values |
| Refusal handling | stop_reason: "refusal" with stop_details; server-side fallbacks can re-run the request on another model, first-party only |
| Tokenizer | Changed between generations — re-baseline token counts after migrating |
Do not assert a capability from memory in production code: client.models.retrieve(id) reports max_input_tokens, max_tokens, and capabilities at runtime.
Breaking behaviour changes across releases
A model upgrade is a breaking-change review, not a string swap.
| Change class | What to check |
|---|---|
| Removed parameters | Sampling parameters, budget_tokens, prefill — a request that worked last year now 400s |
| Changed defaults | Thinking on/off when omitted; thinking display default; effort default |
| New stop reasons | refusal must be handled before reading content |
| Tokenizer change | Cost models, chunk sizes, and truncation thresholds all shift |
| Prompt sensitivity | Prompts written for an older model are often too prescriptive and reduce quality; over-specified scaffolding should be deleted, not extended |
| Tool-call formatting | JSON escaping in tool_use.input varies — parse, never string-match |
| Latency and turn length | Newer models on hard tasks can run much longer per turn; timeouts, streaming, and progress UX need revisiting |
| Cache behaviour | Caches are model-scoped: switching models starts cold |
The migration sequence: run the eval set on the new model before editing prompts → fix hard API breaks → re-baseline tokens → re-tune effort → audit prompts for stale instructions → canary → cut over. Details on 2.2 Systems life cycle.
Quiz
- Question 1
A high-volume classification route runs on an Opus-tier model and its prompts reach 300K tokens. The team wants to move it to Claude Haiku 4.5 for cost. Which fact blocks the move as the route is currently written?
- Question 2
Before building a two-model cascade that routes easy documents to a cheaper model, which measurement comes first?
Links
- https://platform.claude.com/docs/en/about-claude/models/overview
- https://platform.claude.com/docs/en/about-claude/pricing
- https://platform.claude.com/docs/en/about-claude/model-deprecations
- https://platform.claude.com/docs/en/api/models-list
- https://platform.claude.com/docs/en/build-with-claude/extended-thinking