Verified November 2026

Home › Compare › Claude Haiku 4.5 vs Claude Haiku 5.5

Claude Haiku 4.5 vs Claude Haiku 5.5

A straight cost comparison of Claude Haiku 4.5 (Anthropic) and Claude Haiku 5.5 (Anthropic), per 1,000,000 tokens, verified November 2026.

$1Claude Haiku 4.5 input / 1M
$0.10Claude Haiku 5.5 input / 1M
90%Blended price gap
Claude Haiku 5.5Cheaper blended

Rate-by-rate comparison

USD per 1M tokens · November 2026
MetricClaude Haiku 4.5Claude Haiku 5.5Winner
Input / 1M$1$0.10Claude Haiku 5.5
Cached input / 1M$0.10$0.01Claude Haiku 5.5
Cache write / 1M$1.25$0.125Claude Haiku 5.5
Output / 1M$5$0.50Claude Haiku 5.5
Context window200,000200,000Tie
Batch APIYesYesTie
Long-context tierNone> 100,000—

For per-token metrics, the lower number wins. For context window and batch support, the higher or broader option wins. A dash means the provider does not publish that rate: a missing cache-write fee means writes are not billed per token, while a missing cached-input rate means cache hits bill at the full input rate.

Blended cost: 1M input + 250K output

ModelBlended costDifference
Claude Haiku 4.5$2.25+$2.02
Claude Haiku 5.5$0.225Cheaper

Ten workloads, two bills

The same two models, priced across ten workloads that differ only in their input:output ratio. Nothing else changes — no retries, no system prompt, no middleware.

WorkloadClaude Haiku 4.5Claude Haiku 5.5Winner
Classification
100K reqs · 500 in / 10 out
$55$5.50Claude Haiku 5.5
Data extraction
10K reqs · 2K in / 200 out
$30$3Claude Haiku 5.5
Support chat
1K reqs · 4K in / 600 out
$7$0.70Claude Haiku 5.5
Code review
2K reqs · 8K in / 1.5K out
$31$3.10Claude Haiku 5.5
Summarisation
1K reqs · 20K in / 1K out
$25$2.50Claude Haiku 5.5
Agent loop
5K reqs · 12K in / 2K out
$110$11Claude Haiku 5.5
Translation
500 reqs · 15K in / 15K out
$45$4.50Claude Haiku 5.5
Long-document QA
200 reqs · 120K in / 800 out
$24.80$12.08Claude Haiku 5.5
Cached RAG
1K reqs · 50K in (45K cached) / 1K out
$14.55$1.46Claude Haiku 5.5
Nightly batch
10K reqs · 10K in / 2K out
$100$10Claude Haiku 5.5

Claude Haiku 4.5 wins 0 of 10, Claude Haiku 5.5 wins 10 of 10. The winner is set by the token mix, not by the headline rate.

Where the two bills cross over

Claude Haiku 5.5 is cheaper on both sides of the mix: $0.10 vs $1 per million input tokens, and $0.50 vs $5 per million output. No input:output ratio reverses that, so on price alone the other model never becomes the better choice — it has to win on context, latency or output quality.

Cost at four scales

The same support-chat workload (4K input, 600 output per request) at four monthly volumes.

Requests / monthClaude Haiku 4.5Claude Haiku 5.5Difference
100$0.70$0.07Claude Haiku 5.5 cheaper by $0.63
1,000$7$0.70Claude Haiku 5.5 cheaper by $6.30
10,000$70$7Claude Haiku 5.5 cheaper by $63
100,000$700$70Claude Haiku 5.5 cheaper by $630

What caching does to each bill

Caching only helps the part of the prompt that repeats. This is the support-chat workload (1,000 requests, 4,000 input tokens each) with an increasing share of that prompt served from a warm cache, written once on the first request.

Cached share of the promptClaude Haiku 4.5Claude Haiku 5.5Winner
0%$7$0.70Claude Haiku 5.5
50%$5.20$0.52Claude Haiku 5.5
80%$4.12$0.412Claude Haiku 5.5
95%$3.58$0.358Claude Haiku 5.5

Both models publish a cached-read rate, so the discount is comparable: Claude Haiku 4.5 reads cache at $0.10 per million, Claude Haiku 5.5 at $0.01.

Which one should you pick?

Claude Haiku 4.5 wins on none of the ten workloads above, while Claude Haiku 5.5 wins on classification, data extraction, support chat, code review, summarisation, agent loop, translation, long-document qa, cached rag, nightly batch.

Beyond price, the two differ on input rate ($1 vs $0.10); output rate ($5 vs $0.50); cached-input rate ($0.10 vs $0.01); cache-write rate ($1.25 vs $0.125); long-context tier (none vs yes, above 100,000).

Batch mode is available on Claude Haiku 4.5 and Claude Haiku 5.5, which halves the nightly job. Both publish cached-input discounts, so a stable system prompt is worth caching on either.

FAQ

Is Claude Haiku 4.5 cheaper than Claude Haiku 5.5?

On a blended 1M-input-plus-250K-output workload, Claude Haiku 5.5 is cheaper by 90% ($0.225 vs $2.25). The gap changes for cache-heavy prompts and for batch jobs, as the workload table shows.

Which handles longer prompts, Claude Haiku 4.5 or Claude Haiku 5.5?

Both models share the same 200,000-token context window. Watch the long-context threshold on Claude Haiku 5.5: rates rise once input passes it.

Do both models support prompt caching?

Claude Haiku 4.5 bills cached reads at $0.10 per 1M; Claude Haiku 5.5 at $0.01.

Related comparisons

Run your own token volumes