Granite 4.2 3B vs Granite 4.2 8B

Which is better for the money: Granite 4.2 3B by IBM or Granite 4.2 8B by IBM? Intelligence, total benchmark cost, and end-to-end time from Artificial Analysis & other sources, snapshot 2026-10-25.

Trade-off

IBMGranite 4.2 3BOn the open-weights frontier

Intelligence: 9.10
Total cost: $14.9

IBMGranite 4.2 8BOff the frontier

Intelligence: 11.10
Total cost: $53.1

Verdict

Depends on your budget

Granite 4.2 8B scores 2.0 points higher (11.1 vs 9.1) but costs 3.6× as much to run the benchmark ($53.1 vs $14.9; mid-range vs budget cost classes). They are not alternatives at one budget: pick Granite 4.2 8B when you need intelligence above 9.1, and Granite 4.2 3B when 9.1 is enough.

Metric	Granite 4.2 3B	Granite 4.2 8B	Difference
Intelligence (higher is better)	9.1	11.1	+22.0% Granite 4.2 8B is 22.0% more intelligent
Total AA cost (lower is better)	$14.91	$53.07	+255.9% Granite 4.2 8B is 3.6× more expensive
End-to-end time (lower is better)	49.98s	52.08s	+4.2% Granite 4.2 8B is 4.2% slower
Price per 1M tokens (in / out)	$0.03 / $0.12	$0.06 / $0.25
Input modalities	Text	Text

Granite 4.2 8B vs Granite 4.2 3B: 22.0% more intelligence, 255.9% more cost to run the AA benchmark, and 4.2% slower end-to-end.

Open this matchup in the interactive picker Granite 4.2 3B: what beats it Granite 4.2 8B: what beats it

Today's Pareto frontier

Against the frontier

Granite 4.2 3B is off the frontier: at the same or lower cost, GPT-6 Luna (Low) scores 21.50 for $10.6.
Granite 4.2 8B is off the frontier: at the same or lower cost, GPT-6 Luna (High) scores 32.90 for $47.9.

Granite 4.2 3BGranite 4.2 8BToday'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/granite-4-2-3b/granite-4-2-8b · Markdown: /api/v1/compare/granite-4-2-3b/granite-4-2-8b?format=md · llms.txt · agent skill