How does prompt caching help?

Anthropic’s prompt caching stores a prefix of the conversation and bills reused tokens at roughly a tenth of the normal input rate. You mark a cache breakpoint after your system prompt and the stable part of the context. On a 30-turn agent session, most turns reuse that prefix, so the bulk of input tokens becomes cache reads instead of full-price tokens.

How do you trim tool results?

Tool outputs are the silent budget killer. A single git log or directory listing can be thousands of tokens, and it gets re-sent every subsequent turn unless you manage it.

• Truncate large outputs before appending them to history.
• Summarize long results into a few lines once they are no longer needed.
• Drop raw file dumps after the model has acted on them.

In one common harness, this alone halved per-session cost.

What settings bound the bill?

Three config values do the heavy lifting:

1. max_tokens — cap output; Sonnet tends to over-explain, and output is the pricier direction.
2. Max turns — stop the loop at 15-25 rather than letting it run until it “feels done.”
3. Context window budget — track cumulative tokens and compact history past a threshold.

Pair these with the same hybrid routing used elsewhere: keep minimax-m2 as a cheap workhorse for simple turns and call Sonnet only where quality matters.

What is a realistic saving?

An unmanaged 30-turn session on a mid-sized repo can hit 400k-800k input tokens. With caching on the stable prefix and tool-result trimming, the billable fresh input can drop below 100k. On Sonnet’s input rate, that is the difference between a few dollars and well under a dollar for the same work.

What do you give up?

Caching adds a little latency on the first turn and code complexity in your harness. Trimming tool results risks cutting a detail the model later needs — keep the last relevant result intact. Neither change touches model choice, so answer quality on the actual task is unaffected.

Quick Summary

• Prompt caching bills reused prefix tokens at roughly a tenth of input price.
• Trim or summarize large tool outputs before appending them.
• Cap max_tokens, set 15-25 max turns, compact long history.
• A 30-turn session can drop from 400k-800k fresh input tokens to under 100k.
• Adds modest harness complexity; task quality stays the same.

Last updated September 2026.

What is Claude Sonnet 4.5?

Sonnet 4.5 is a general-purpose model from Anthropic with strong tool-calling and code generation. It supports a large context window, accepts images, and exposes the same messages API as other Claude models. In the Claude lineup it is the pragmatic middle: more capable than Haiku, cheaper than Opus.

How do you connect it to an agent?

The wiring is minimal:

1. Take an Anthropic API key and export it as ANTHROPIC_API_KEY.
2. Point your agent at the model string for Sonnet 4.5.
3. Define your tools in the standard schema — Sonnet handles parallel tool calls well.
4. Set a max output token cap to keep responses bounded.

Because the API is stable across Claude versions, switching from Haiku or Opus is a one-line edit. Tools like droid let you swap the backing model in config without reinstalling.

How does Sonnet 4.5 price against Opus?

Sonnet costs a fraction of Opus per million tokens — roughly one-fifth on input. On a typical coding task burning 50k tokens, Sonnet runs a few cents while Opus runs closer to twenty. For 90% of day-to-day edits the quality gap does not show up, which makes Sonnet the sensible default and Opus the escalation path.

When should you still pick Opus?

Reach for Opus on architecture decisions, subtle concurrency bugs, and large refactors where a wrong plan wastes more than the token premium. A good split: Sonnet writes the change, Opus reviews the plan on high-stakes tasks. This mirrors the hybrid routing that keeps agent costs sane.

What are the limits?

Sonnet can over-narrate — it explains more than needed, which inflates output tokens. It also sometimes edits more files than requested unless you scope the task. Both are prompt-fixable: ask for the diff only, and name the target files explicitly.

Quick Summary

• Claude Sonnet 4.5 is Anthropic’s mid-tier coding model, below Opus on cost.
• Connect via the Anthropic API; swapping models is a one-line change.
• Costs roughly one-fifth of Opus on input tokens.
• Best default for routine edits; escalate to Opus for hard planning.
• Scope tasks tightly — Sonnet can over-narrate and over-edit.

Last updated September 2026.