Model comparison
Gemma 4 26B A4B IT vs Granite 3.0 8b Instruct
Gemma 4 26B A4B IT is the stronger model overall, scoring 43.5 to 31.6 on the Noometry Index.
Last verified . 12 shared benchmarks.
GoogleSummary
- They share 12 benchmarks with published results for both. Gemma 4 26B A4B IT scores higher in 8 categories and Granite 3.0 8b Instruct in 0 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in writing & preference, where Gemma 4 26B A4B IT leads 58.6 to 31.1.
Side by side
| Gemma 4 26B A4B IT | Granite 3.0 8b Instruct | |
|---|---|---|
| Provider | IBM | |
| Noometry Index | 43.5 | 31.6 |
| Released | 2026-04-02 | — |
| Weights | Open | Open |
| Context window | 262K | — |
| Max output | 33K | — |
| Input $ / M tokens | $0.09 | — |
| Output $ / M tokens | $0.30 | — |
| Results tracked | 28 | 14 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Gemma 4 26B A4B IT leads
Gemma 4 26B A4B IT: 39.0 (#164), Granite 3.0 8b Instruct: 29.7 (#301)
| Benchmark | Gemma 4 26B A4B IT | Granite 3.0 8b Instruct |
|---|---|---|
| LMArena Coding | 1447 | 1112 |
| LMArena WebDev | 1359 | — |
| SciCode | 40% | — |
| WeirdML | 35.2% | — |
| BigCodeBench Instruct | — | 29.3% |
| BigCodeBench Complete | — | 35.4% |
| ALE-Bench | 927.17 | — |
Reasoning Too close to call
Gemma 4 26B A4B IT: 21.8 (#213), Granite 3.0 8b Instruct: 20.9 (#230)
| Benchmark | Gemma 4 26B A4B IT | Granite 3.0 8b Instruct |
|---|---|---|
| LMArena Hard Prompts | 1439 | 1092 |
| CritPt | 0% | — |
| Chess Puzzles | 6% | — |
| DTBench | 74.9% | — |
| LMCA | 29.7% | — |
| Epoch Capabilities Index | 141.85 | — |
Math Gemma 4 26B A4B IT leads
Gemma 4 26B A4B IT: 47.6 (#67), Granite 3.0 8b Instruct: 32.8 (#210)
| Benchmark | Gemma 4 26B A4B IT | Granite 3.0 8b Instruct |
|---|---|---|
| LMArena Math | 1470 | 1143 |
| OTIS Mock AIME 2024-2025 | 82.2% | — |
Knowledge Gemma 4 26B A4B IT leads
Gemma 4 26B A4B IT: 45.7 (#85), Granite 3.0 8b Instruct: 29.6 (#236)
| Benchmark | Gemma 4 26B A4B IT | Granite 3.0 8b Instruct |
|---|---|---|
| LMArena Expert | 1447 | 1087 |
| GPQA Diamond | 73.2% | — |
| Vectara Hallucination Rate | 5.2% | — |
Multimodal Not comparable
Gemma 4 26B A4B IT: 40.6 (#46), Granite 3.0 8b Instruct: —
| Benchmark | Gemma 4 26B A4B IT | Granite 3.0 8b Instruct |
|---|---|---|
| LMArena Vision | 1260 | — |
Multilingual Gemma 4 26B A4B IT leads
Gemma 4 26B A4B IT: 53.1 (#71), Granite 3.0 8b Instruct: 27.2 (#276)
| Benchmark | Gemma 4 26B A4B IT | Granite 3.0 8b Instruct |
|---|---|---|
| LMArena Non-English | 1421 | 1037 |
| LMArena Chinese | 1495 | 1063 |
| LMArena Russian | 1434 | 1060 |
| LMArena French | 1460 | — |
| LMArena Spanish | 1417 | — |
Instruction Following Gemma 4 26B A4B IT leads
Gemma 4 26B A4B IT: 74.8 (#82), Granite 3.0 8b Instruct: 56.0 (#276)
| Benchmark | Gemma 4 26B A4B IT | Granite 3.0 8b Instruct |
|---|---|---|
| LMArena Instruction Following | 1420 | 1088 |
Long Context Gemma 4 26B A4B IT leads
Gemma 4 26B A4B IT: 43.6 (#91), Granite 3.0 8b Instruct: 34.0 (#252)
| Benchmark | Gemma 4 26B A4B IT | Granite 3.0 8b Instruct |
|---|---|---|
| LMArena Longer Query | 1428 | 1122 |
Writing & Preference Gemma 4 26B A4B IT leads
Gemma 4 26B A4B IT: 58.6 (#115), Granite 3.0 8b Instruct: 31.1 (#285)
| Benchmark | Gemma 4 26B A4B IT | Granite 3.0 8b Instruct |
|---|---|---|
| LMArena Text | 1434 | 1096 |
| LMArena Creative Writing | 1402 | 1071 |
| LMArena Multi-Turn | 1441 | 1063 |
| EQ-Bench Creative Writing | 1305 | — |
Frequently asked questions
Is Gemma 4 26B A4B IT better than Granite 3.0 8b Instruct?
Gemma 4 26B A4B IT is the stronger model overall, scoring 43.5 to 31.6 on the Noometry Index.
Is Gemma 4 26B A4B IT or Granite 3.0 8b Instruct better for coding?
Gemma 4 26B A4B IT scores higher on coding benchmarks: 39.0 versus 29.7 in the Noometry coding category.
How many benchmarks do Gemma 4 26B A4B IT and Granite 3.0 8b Instruct share?
12 benchmarks have published results for both models. Gemma 4 26B A4B IT has 28 scored results on Noometry and Granite 3.0 8b Instruct has 14.