Claude Haiku 4.5 API pricing
Published on-demand rates for Anthropic's claude-haiku-4-5, captured 2026-09-21 from public pricing data and committed to our archive. Context window 200k tokens; first observed in our archive 2025-10-20.
| Meter | USD per 1M tokens |
|---|---|
| Input | $1.00 |
| Output | $5.00 |
| Cached input | $0.100 |
Standard on-demand rates. Excludes batch endpoints and negotiated pricing; confirm on Anthropic's pricing page before committing.
What it costs on a real workload
Per 1M tokens processed, at the rates above. How the four profiles are defined.
| Profile | In / out mix | Cached share | Cost |
|---|---|---|---|
| Retrieval | 800k / 200k | 70% | $1.30 |
| Chat | 500k / 500k | 30% | $2.87 |
| Content | 200k / 800k | 20% | $4.16 |
| Agent | 900k / 100k | 85% | $0.712 |
Observed rate history
No repricing observed since this model entered our archive on 2025-10-20. When a rate moves, the change appears here and on the sitewide ticker.
What the record shows
Claude Haiku 4.5 has not repriced once since it entered our archive on 2025-10-20. That is the ordinary case rather than the exception: providers ship new models at lower rates far more often than they cut the price of one already in service, which is why a cheaper option usually arrives as a new name rather than as a smaller number on this page.
Against the 40 models we price, it is the 24th cheapest for retrieval-shaped work, which reads a great deal and writes little, and the 24th cheapest for content-shaped work, which does the reverse. The two are close, so it holds its position whichever way your workload leans.
It charges 5.0 times more for output than for input, against a median of 5.0 times across the set. That is close to typical for the models we track.
Inside Anthropic's own lineup it is the cheapest of the 4 models we track on this workload, which makes it the floor to measure the rest of the house against before looking outside it at all.
One other model we price charges exactly $1.00 in and $5.00 out: Sonar Reasoning. Where a rate ties exactly, price has stopped being the deciding factor and this site has nothing further to tell you: that choice comes down to your own evaluations, which we do not run and do not score.
Cached input reads at $0.100 against a standard input rate of $1.00, a 90% discount on the part of your prompt that repeats.
It accepts image input and Anthropic publishes a tokenisation rule for it, so a scanned page can be priced honestly: about 1,540 tokens for a US Letter page at 200 DPI, before the instructions you send with it.
Its 200k token context window holds roughly 150,000 words of English text in a single request, or about 129 scanned pages at the per-page figure above. A workload that does not fit cannot run here at any price, which is a capability limit rather than a cost one.
Priced near this one
Closest to Claude Haiku 4.5 on a retrieval workload, cheaper and more expensive alike. Cost only; we publish no quality score.
| Model | Provider | Input | Output | vs Claude Haiku 4.5 |
|---|---|---|---|---|
| DeepSeek V4 Pro | DeepSeek | $1.32 | $3.96 | 13% less |
| GPT 5.4 Mini | OpenAI | $0.750 | $4.50 | 13% less |
| Sonar | Perplexity | $1.00 | $1.00 | 23% less |
| Gemini 3.8 Flash | $0.750 | $3.75 | 25% less |
Retrieval profile: 800k in, 200k out, 70% cached where offered.
No affiliate or referral arrangements: the console link carries no parameters and nothing on this page is paid placement. Claude Haiku 4.5 is a trademark of its owner; LLMPrice.io is independent and unaffiliated.