Home · Product

Claude Haiku batch API for CSV jobs

Model catalogs stop at benchmarks. BatchHaiku is the CSV → Haiku → CSV path: templates, metering, export, and a REST API on Claude Haiku 5.5 via Cloudflare AI.

What you can run

Classify

Reviews, tickets, leads — frozen taxonomy, one label per row.

Extract

Dates, amounts, names, intents → compact JSON or CSV columns.

Filter

RAG chunk keep/drop before embed — shared system prompt caches well.

Label

Build training / QA sets without standing up a DIY pipeline.

Limits & metering

UI auto-splits large CSVs; each upstream call is capped at 50 rows for latency.
~1 credit per short labeled row — see credit packs.
Live model: Claude Haiku 5.5 (falls back to Haiku 4.5 if routing fails).
Anthropic list rates for budgeting: Claude Haiku pricing.

REST example

Authenticated batch endpoint (see API docs). Shape is illustrative:

curl -X POST https://batchhaiku.dev/api/v1/batch/ \
  -H "Authorization: Bearer YOUR_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "templateId": "review_sentiment",
    "rows": ["Love this app", "Crashes on launch"]
  }'

Cost tips for Haiku 5.5

Short system prompts; strip unused CSV columns.
Validate 20–50 gold rows before full dump.
Leave room in max_tokens when adaptive thinking is on.
Prefer Haiku over Sonnet for pure tagging — Haiku vs Sonnet.

BatchHaiku is an independent third-party tool, NOT affiliated with or endorsed by Anthropic. Live inference runs Claude Haiku 5.5 via Cloudflare AI under your account usage terms.

Open batch tool API docs

Related guides

Claude Haiku pricing (4.5 & 5.5)
List rates, long-context tiers, and cost per 1k CSV rows.
Claude Haiku 5.5 benchmarks
AA Index, Terminal-Bench, SWE — with sources cited.
Haiku vs Sonnet for batch
When Haiku is enough and when Sonnet is worth the spend.
Haiku 5.5 vs GPT-6 Luna
Same short-tier price — different long-context thresholds & scores.
Live model comparison tool
Switch Haiku, Luna, Sonnet, Opus side by side on the homepage.