The AI arena is free today

Open Superagent

IBM Granite 4.2 30B vs IBM Granite 4.2 8B

IBM Granite 4.2 30B and IBM Granite 4.2 8B are closely matched at 24.0 and 19.9 on the LLM Stats Score.

IBM · IBM · Updated for 2026

Which is better?

IBM Granite 4.2 30B and IBM Granite 4.2 8B are closely matched on the overall LLM Stats Score at 24.0 and 19.9.

In the 18 individual benchmarks reported for both models, IBM Granite 4.2 30B wins 17; this is a narrower head-to-head signal than the composite indexes.

Based on current LLM Stats indexes, shared benchmarks, pricing, and model metadata for 2026.

Choose IBM Granite 4.2 30B

  • you value its reported benchmark strengths — it wins 17 of 18 exact shared results

Choose IBM Granite 4.2 8B

  • you are already invested in the IBM ecosystem

At a glance

The differences that matter most.

Core performance indexes
24.0
#156
19.9
#185
21.8
#165
17.6
#193
9.6
#159
2.8
#210
7.9
#124
2.6
#150
Cost, coverage & limits
Benchmark wins
17 of 18
1 of 18
Input price
— / M
— / M
Output price
— / M
— / M
Context window

Capability indexes

Additional strengths measured across groups of related public benchmarks

3 shared
Index
IBM Granite 4.2 30B
IBM Granite 4.2 8B
25.6#103
21.2#141
10.6#116
4.9#145
14.0#50
7.8#86
Conservative TrueSkill rating · higher is betterHow scores work

Individual benchmarks

18 reported for IBM Granite 4.2 30B · 18 for IBM Granite 4.2 8B

18 shared

IBM Granite 4.2 30B outperforms in 17 benchmarks (AIME 2025, Arena-Hard v2, BFCL-V4, Bird-SQL (dev), GPQA, HMMT25, LiveCodeBench v6, MMLU-Pro, ProfBench, RULER 128k, RULER 64k, SciCode, SWE-bench Multilingual, SWE-Bench Pro, SWE-Bench Verified, TAU3-Bench, Terminal-Bench 2.1), while IBM Granite 4.2 8B is better at 1 benchmark (IFBench).

IBM Granite 4.2 30B significantly outperforms across most benchmarks.

Mon Aug 31 2026 • llm-stats.com

Human preference

Blind head-to-head votes and playground preference scores

Model Size

Parameter count comparison

22.0B diff

IBM Granite 4.2 30B has 22.0B more parameters than IBM Granite 4.2 8B, making it 275.0% larger.

IBM
IBM Granite 4.2 30B
30.0Bparameters
IBM
IBM Granite 4.2 8B
8.0Bparameters
30.0B
IBM Granite 4.2 30B
8.0B
IBM Granite 4.2 8B

License

Usage and distribution terms

Both models are licensed under Apache 2.0.

Both models share the same licensing terms, providing consistent usage rights.

IBM Granite 4.2 30B

Apache 2.0

Open weights

IBM Granite 4.2 8B

Apache 2.0

Open weights

Release Timeline

When each model was launched

Both models were released on 2026-08-25.

They likely represent similar generations of model development.

IBM Granite 4.2 30B

Aug 25, 2026

6 days ago

IBM Granite 4.2 8B

Aug 25, 2026

6 days ago

Knowledge Cutoff

When training data ends

Neither model specifies a knowledge cutoff date.

Unable to compare the recency of their training data.

No cutoff dates available

Outputs Comparison

Notice missing or incorrect data?Start an Issue discussion

Judge for yourself.

Run your own prompts against IBM Granite 4.2 30B and IBM Granite 4.2 8B side-by-side, then vote on the output you prefer.

IBM Granite 4.2 30B
✓ Preferred
IBM Granite 4.2 8B
Open in Playground

FAQ

Common questions about IBM Granite 4.2 30B vs IBM Granite 4.2 8B.

Which is better, IBM Granite 4.2 30B or IBM Granite 4.2 8B?

IBM Granite 4.2 30B and IBM Granite 4.2 8B are closely matched on the LLM Stats Score at 24.0 and 19.9. IBM Granite 4.2 30B is made by IBM and IBM Granite 4.2 8B is made by IBM. The best choice depends on your use case — compare their capability indexes, individual benchmarks, pricing, and limits above.

How does IBM Granite 4.2 30B compare to IBM Granite 4.2 8B in benchmarks?

IBM Granite 4.2 30B scores RULER 64k: 90.0%, AIME 2025: 89.2%, HMMT25: 89.2%, RULER 128k: 81.4%, MMLU-Pro: 77.6%. IBM Granite 4.2 8B scores AIME 2025: 86.7%, RULER 64k: 81.0%, IFBench: 79.3%, HMMT25: 78.3%, MMLU-Pro: 74.0%.

What are the main differences between IBM Granite 4.2 30B and IBM Granite 4.2 8B?

Key differences include LLM Stats Score (24.0 vs 19.9). See the full comparison above for benchmark-by-benchmark results.