BatchHaiku

Home · Model comparisons

Claude Haiku 5.5 vs GPT-6 Luna

Two low-cost models with the same short-context sticker price — different long-prompt tiers, benchmarks, and token spend at max effort. Use this page for the comparison intent; use BatchHaiku home for Haiku-first batch positioning.

At a glance

AI model comparison

Haiku 5.5 vs Luna 6

Independent + vendor-reported scores — harnesses differ; compare within a row.

Metric
Haiku 5.5
Luna 6

Intelligence Index (max)

Artificial Analysis

43

38

Terminal-Bench 4.0

AA protocol (vendor cards may differ)

33%

13%

AutomationBench-AA

~35%

~53%

Best for

Haiku 5.5 — Bulk classify, extract, label, agent subtasks

Luna 6 — Long-prompt cheap tier, lighter reasoning spend

Quick read

Haiku 5.5 leads Intelligence Index (43 vs 38). Short-tier list prices match — long-context thresholds decide the bill. Haiku 5.5 streams faster at max effort (~237 tok/s AA).

Sources: Anthropic overview + Artificial Analysis (max effort); OpenAI pricing notes + Artificial Analysis (max effort). Scores and prices move — verify live provider cards before procurement. BatchHaiku runs live jobs on Claude Haiku via Cloudflare AI.

How to decide for your CSV

Ignore headline screenshots. Measure label accuracy on a gold set, p95 latency, and fully loaded $/1k rows — including long-context tier flips. For Anthropic-side rates see Claude Haiku pricing; for scores see Haiku 5.5 benchmarks.

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.

Run bulk jobs on Haiku 5.5

Related guides