Mid-tier
US$3.00
Claude Sonnet 4.6
Anthropic
Anthropic · Budget · US$1.00 input/1M · US$5.00 output/1M · 200K context
Typical monthly cost
US$20.07
≈ per day
US$0.91
Blended cost/1M
US$1.22
Context window
200K
Claude Haiku 4.5 from Anthropic is a budget model priced at US$1.00 per 1M input tokens and US$5.00 per 1M output tokens. For a typical solo-developer workload (8 hours/day, 22 days/month — 1 medium feature, 5 small bug fixes, 4 PR reviews, 2 stack-trace debugs, ~1500 lines of TypeScript, 1 large-doc read, with prompt caching at the default mix) Claude Haiku 4.5 costs about US$20/month. The 200K-token context window covers single-file workloads comfortably.
Move the slider or switch task mix — values update live.
Monthly budget
US$100 / month
≈ $23/wk · ≈ $4.55/day
on Claude Haiku 4.5
Input tokens
154.5M
Output tokens
3.0M
Total tokens
157.5M
Per month this budget delivers
At the default coding-agent mix with 50% cache hits.
The 22-day month is based on the median working-day count across DE/US.
| Activity | Count | Per task | Daily | Monthly |
|---|---|---|---|---|
| Medium feature (10–15 files) | 1 | US$0.35 | US$0.35 | US$7.67 |
| Small bug fix | 5 | US$0.02 | US$0.11 | US$2.48 |
| PR review | 4 | US$0.03 | US$0.11 | US$2.33 |
| Debug from stack trace | 2 | US$0.05 | US$0.09 | US$2.00 |
| Read a large doc | 1 | US$0.04 | US$0.04 | US$0.77 |
| Micro-interaction (explain / lint fix) | 30 | US$0.00 | US$0.05 | US$1.02 |
| Lines of TypeScript | 1,500 | US$0.00 | US$0.17 | US$3.80 |
| Total | US$0.91 | US$20.07 | ||
The 1500-lines-of-TS row models ~1000 lines read (cache-hit) + ~500 lines written. Headline figures are precise to ~5% — see the FAQ.
What each monthly budget buys on this model (typical solo-developer day, 22 working days).
| Monthly budget | Medium features | PR reviews | Debug sessions | Lines of TS |
|---|---|---|---|---|
| Typical (≈ $20) | 57 | 757 | 441 | 174,517 |
| $50/month | 143 | 1,886 | 1,098 | 434,782 |
| $200/month | 573 | 7,547 | 4,395 | 1,739,130 |
| $500/month | 1,433 | 18,867 | 10,989 | 4,347,826 |
| $2000/month | 5,734 | 75,471 | 43,956 | 17,391,304 |
Typical mix: coding-agent (85% input, 50% cache hits). Values show the maximum count of each task type at that budget.
Coding
Trained or post-trained for code generation tasks.
Reasoning
Strong multi-step reasoning over complex prompts.
Multimodal
Accepts images alongside text.
Prompt cache
Cache reads billed at ~10% of input price — cuts agent costs sharply.
Batch API
50% off when you accept up to 24-hour turnaround.
Tool use
Native function-calling / tool-use API support.
Long context
Not supported
Extended thinking
Not supported
Total models
5
Median input/1M
US$3.00
Median output/1M
US$15.00
Input range
US$1.00–US$15.00
Verified: 2026-05-07
Running the realistic solo-developer day (1 medium feature + 5 small bug fixes + 4 PR reviews + 2 debug sessions + ~1500 lines of TypeScript + 1 large-doc read, 22 working days) on Claude Haiku 4.5 costs about US$20/month. Heavier workloads scale proportionally; lighter workloads cost less.
200K tokens total, with up to 8K of output. That fits a few dozen source files in a single call.
Providers charge US$5.00 per 1M output tokens against US$1.00 per 1M input — output requires real compute, input comes mostly from cache. Coding agents read many files (input-heavy) and emit compact diffs (low output), so total spend is usually input-driven.
Cache reads typically cost only 10% of the regular input rate. On a coding-agent mix with 50% cache hits, that saves roughly 45% on input — which is about 38% off your total bill on input-heavy workloads. Anthropic models charge a one-time cache-write surcharge (25% over input) that pays for itself after 2–3 hits.
Yes, if you can tolerate up to 24-hour turnaround: batch input/output are 50% cheaper than real-time rates. Perfect for nightly code reviews, bulk refactors or pre-merge analysis — wrong for inner-loop editing where you need an answer in seconds.
Open the full calculator with your own budget, task mix and region (US or DE with 19% VAT).
Open calculator