Claude Sonnet 4.6 has consolidated as the default model for most production workloads in 2026. More capable than Haiku, more economical than Opus, with reasonable latency. After three months of intensive use across projects, patterns where it wins and loses are clear.

Key takeaways

  • Sonnet 4.6 covers 80% of production traffic with quality indistinguishable from Opus in blind tests.

  • Token cost for Sonnet 4.6 is around 60% of Opus 4.7’s, per Anthropic’s official pricing[1].

  • Complex multi-step reasoning, agentic coding over large codebases, and multi-thread analysis still need Opus.

  • Dynamic router (Haiku as classifier + Sonnet/Opus by complexity) lowers average cost 40-60% versus "all Sonnet".

  • Empirical calibration is the only reliable way to decide when to escalate to Opus.

Where Sonnet 4.6 suffices (80% of traffic)

Tasks where Sonnet 4.6 produces quality indistinguishable from Opus in blind tests, at a token cost around 60% of Opus’s:

  • Classification.

  • Structured extraction.

  • Summarisation.

  • Support drafting.

  • First-response agent.

  • Medium-complexity code generation.

The usual pattern is routing 70-80% of traffic to Sonnet and reserving Opus for what actually needs it. Teams using Opus by default "to be safe" pay about 67% more per token on tasks where Sonnet already suffices, with no measurable gain.

Where Sonnet falls short

Tasks where Opus 4.7 still leads:

  • Complex multi-step reasoning.

  • Agentic coding over large codebases.

  • Analysis requiring many simultaneous threads.

  • Strategic decisions with multiple trade-offs.

On these tasks, Sonnet’s savings don’t offset the cost of a mediocre response.

Detection is empirical: same task with Sonnet and Opus, rubric evaluation by human or LLM-as-judge:

  • Gap greater than one point on 5-scale → use Opus.

  • Gap under half a point → Sonnet is enough.

Dynamic router as norm

The stack we see working best in 2026 is a three-tier inference router:

  • Haiku 4.5 as classifier: cheap, fast, classifies queries by expected complexity.

  • Sonnet 4.6 for 70-80% of queries.

  • Opus 4.7 for queries exceeding the complexity threshold.

With decent calibration, the resulting mix has 40-60% lower average cost than "all Sonnet" with aggregate quality indistinguishable. If your team already juggles more than one provider, an LLM proxy like LiteLLM makes it easier to centralize that routing without rewriting every integration.

What the classifier should look at

The router is only as good as the signal it routes on. In practice a cheap classifier decides the tier from a handful of features, not a deep read of the query:

  • Steps implied. A single extraction or reply stays on Sonnet; anything that reads like "plan, then do, then verify" is an Opus signal.

  • Context size and spread. Answers grounded in one document stay low; questions that force the model to hold more than one source or file at once escalate.

  • Tool and branch count. Zero or one tool call is routine; multi-tool, multi-branch agentic loops are where Sonnet starts to drift and Opus earns its price.

  • Cost of being wrong. A throwaway summary and an irreversible migration deserve different tiers even at identical complexity: route on blast radius, not just difficulty.

Start every one of these on Sonnet and let the classifier promote to Opus on a positive signal, rather than defaulting high and hoping to save later. Re-calibrate the thresholds monthly against a small rubric-scored sample: query mix drifts, and a router tuned six months ago is quietly overpaying or under-serving today.

Conclusion

Sonnet 4.6 is the 2026 workhorse for a reason: it’s the default choice because the capability-cost-latency balance is the best on the market. Using it as default with a router escalating to Opus when needed is the reference architecture in mature implementations. Teams still using Opus by default for all tasks are paying a tax that doesn’t buy extra quality.

This article is also available in Spanish: Claude Sonnet 4.6 en producción: el equilibrio coste-calidad.

Sources:

  1. Anthropic: models overview and pricing (Claude Sonnet 4.6, Claude Opus 4.7)[1]
  2. Claude: Pricing[2]
  3. Anthropic: Introducing Claude 4[3]

Frequently asked questions

How much do I save by using Sonnet 4.6 instead of Opus 4.7?

Sonnet 4.6's token cost is around 60% of Opus 4.7's per Anthropic's official pricing. A team using Opus by default therefore pays about 67% more per token on tasks where Sonnet already suffices, with no measurable gain. On classification, structured extraction, summarisation, support drafting, first-response agents and medium-complexity code generation, quality is indistinguishable from Opus in blind tests.

How do I know whether a given task needs Opus or Sonnet is enough?

Empirically: run the same task with Sonnet and with Opus and evaluate with a rubric, by a human or with LLM-as-judge. If the gap is greater than one point on a 5-point scale, use Opus; if it is under half a point, Sonnet is enough. The tasks where Opus remains superior are complex multi-step reasoning, agentic coding over large codebases, analysis requiring simultaneous threads and strategic decisions with multiple trade-offs.

Which signals should the classifier in a three-tier router use?

Four cheap-to-detect features: the steps implied (plan, then do, then verify is an Opus signal) and context size and spread. Also weigh the tool and branch count of the agentic loop, and the cost of being wrong, routing on blast radius rather than difficulty alone. With Haiku 4.5 as classifier, Sonnet 4.6 for 70-80% of queries and Opus 4.7 for the rest, average cost drops 40-60%. Re-calibrate the thresholds monthly.

Sources

  1. Anthropic’s official pricing
  2. Claude: Pricing
  3. Anthropic: Introducing Claude 4