Model Tiers & the Tradeoff Triangle
CoreSelect Claude models and apply model-choice tradeoffs · Difficulty 2/5
Explanation
Model Tiers
| Tier | Character | Typical fit |
|---|---|---|
| Haiku | Fastest, cheapest | High-volume, well-defined, latency-sensitive steps (classification, routing, extraction) |
| Sonnet | Balanced capability/cost | The workhorse for most production tasks |
| Opus | Most capable | The hardest reasoning, planning, and orchestration steps |
Right-size the model to the task rather than defaulting to the most capable tier available.
The Tradeoff Triangle
Quality, latency, and cost trade against each other. There is no universally "best" model -- only the model that fits a specific requirement. An architect resolves model choice against the constraint (e.g., a sub-second latency SLA, or a budget per 1,000 requests), not in the abstract.
Mixed-Model Designs
Production pipelines rarely use one model throughout. Mixed-model designs route cheap, well-defined steps (classification, routing, simple extraction) to Haiku, and hard reasoning, planning, or orchestration steps to Sonnet or Opus. This mirrors multi-agent orchestration patterns, where a coordinator (often a stronger model) delegates narrow subtasks to cheaper workers.
Common exam traps
- "Always pick the most capable model." This wastes cost and latency on tasks a smaller tier handles fine. The correct answer fits the constraint, not the ceiling.
- Ignoring that model choice is a per-step decision in a pipeline -- a single "best" model for the whole system is often the wrong frame.
Key Takeaways
- Haiku = fastest/cheapest for high-volume, well-defined steps; Sonnet = balanced workhorse; Opus = most capable for hard reasoning/orchestration
- Resolve quality vs. latency vs. cost against the specific requirement, not in the abstract
- Mixed-model pipelines route each step to the tier that fits its difficulty and budget
- Defaulting to the most capable model everywhere is a common exam trap, not a best practice
Glossary Terms
The most capable Claude model tier, excelling at complex multi-step reasoning, nuanced analysis, and high-stakes creative tasks. Highest accuracy but most expensive and slowest. Best suited for tasks where output quality outweighs cost and latency concerns.
The balanced Claude model tier offering the best trade-off between capability, speed, and cost for most production applications. The recommended default for new systems — handles the majority of complex tasks effectively without Opus-level expense.
The fastest and most cost-effective Claude model tier, optimized for high-throughput, low-latency tasks like classification, extraction, and simple Q&A. Carries a 200K-token context limit (vs 1M on Sonnet/Opus) and is the recommended first-pass router in tiered pipeline architectures.
The practice of directing requests to different Claude model tiers based on assessed complexity and requirements. A common pattern uses a fast, cheap model (Haiku) to classify task complexity, then routes to Sonnet or Opus accordingly.
Related Concepts