Claude Haiku 5.5 (Medium) vs Granite 4.2 3B
Which is better for the money: Claude Haiku 5.5 (Medium) by Anthropic or Granite 4.2 3B by IBM? Intelligence, total benchmark cost, and end-to-end time from Artificial Analysis & other sources, snapshot 2026-10-24.
Trade-off
Verdict
Depends on your budget
Claude Haiku 5.5 (Medium) scores 25.4 points higher (34.5 vs 9.1) but costs 3.8× as much to run the benchmark ($56.4 vs $14.9; mid-range vs budget cost classes). They are not alternatives at one budget: pick Claude Haiku 5.5 (Medium) when you need intelligence above 9.1, and Granite 4.2 3B when 9.1 is enough.
| Metric | Claude Haiku 5.5 (Medium) | Granite 4.2 3B | Difference |
|---|---|---|---|
| Intelligence (higher is better) | 34.5 | 9.1 | −73.6% Granite 4.2 3B is 73.6% less intelligent |
| Total AA cost (lower is better) | $56.36 | $14.91 | −73.5% Granite 4.2 3B is 73.5% cheaper |
| End-to-end time (lower is better) | 10.41s | 35.30s | +239.1% Granite 4.2 3B is 239.1% slower |
| Price per 1M tokens (in / out) | $0.10 / $0.50 | $0.03 / $0.12 | |
| Input modalities | Text, Image | Text |
Granite 4.2 3B vs Claude Haiku 5.5 (Medium): 73.6% less intelligence, 73.5% less cost to run the AA benchmark, and 239.1% slower end-to-end.
Today's Pareto frontier
Against the frontier
- Claude Haiku 5.5 (Medium) is on today's frontier: nothing is both smarter and cheaper.
- Granite 4.2 3B is off the frontier: at the same or lower cost, GPT-6 Luna (Low) scores 21.50 for $10.6.
Claude Haiku 5.5 (Medium)Granite 4.2 3BToday's Pareto frontierOpen-weights frontierEvery other model
Matchup
Compare something else
Agents:
Accept: text/markdown or ?format=md returns this page as Markdown. JSON: /api/v1/compare/claude-haiku-5-5-medium/granite-4-2-3b · Markdown: /api/v1/compare/claude-haiku-5-5-medium/granite-4-2-3b?format=md · llms.txt · agent skill