Language modelActive
Granite 4.1 8B
IBM
- Released
- April 29, 2026
- Data date
- September 8, 2026
- Speed
- 58.1 tokens/sOutput throughput
- Price
- Input · Output / 1M tokens
- Input
- Text
- Output
- Text
Granite 4.1 8B remains a dense 8B transformer despite its 131,072-token context. IBM documents fill-in-the-middle and lists twelve supported languages including German, English, French, Japanese, and Chinese.
The model card describes GQA, RoPE, SwiGLU, and RMSNorm as core architectural pieces, while also warning that multilingual performance can differ from English across languages. Apache 2.0 permits reuse of the weights.
Specifications and access
| Specification | Value and source |
|---|---|
| Weights | Open weightsSource |
| Access | API, LocalSource |
| Parameters | 8BSource |
| Context window | 131,072 tokensSource |
| License | Apache-2.0Source |
| Release | 2026-04-29Source |
| Parameters | 8 BSource |
| Architecture | dense decoder-only transformer with GQA, RoPE, SwiGLU, RMSNorm and shared embeddingsSource |
| Input | TextSource |
| Output | TextSource |
| Languages | English, German, Spanish, French, Japanese, Portuguese, Arabic, Czech, Italian, Korean, Dutch, ChineseSource |
| Tool calling | YesSource |
Page 1 of 4
Measurements without matching peer values
These measurements have no matching peer values under the same test conditions. Their original values and sources remain available here.
MMLU (Show measurement, test conditions, and source)
- Source value
- 73.84
- Score
- 73.84
- Metric
- percent
- Unit
- %
- Category
- mixed
- Direction
- Higher is better
- Source type
- vendor-reported
- Evaluator
- IBM Granite 4.1 8B model card
- Status
- active
- Retrieved at
- 2026-09-08
- Methodology
- Vendor-reported result. Vendor evaluation table; row-specific shot/CoT and task settings are retained in the card table.
MMLU-Pro (Show measurement, test conditions, and source)
- Source value
- 55.99
- Score
- 55.99
- Metric
- percent
- Unit
- %
- Category
- mixed
- Direction
- Higher is better
- Source type
- vendor-reported
- Evaluator
- IBM Granite 4.1 8B model card
- Status
- active
- Retrieved at
- 2026-09-08
- Methodology
- Vendor-reported result. Vendor evaluation table; row-specific shot/CoT and task settings are retained in the card table.
BBH (Show measurement, test conditions, and source)
- Source value
- 80.51
- Score
- 80.51
- Metric
- percent
- Unit
- %
- Category
- mixed
- Direction
- Higher is better
- Source type
- vendor-reported
- Evaluator
- IBM Granite 4.1 8B model card
- Status
- active
- Retrieved at
- 2026-09-08
- Methodology
- Vendor-reported result. Vendor evaluation table; row-specific shot/CoT and task settings are retained in the card table.
AGI EVAL (Show measurement, test conditions, and source)
- Source value
- 72.43
- Score
- 72.43
- Metric
- percent
- Unit
- %
- Category
- mixed
- Direction
- Higher is better
- Source type
- vendor-reported
- Evaluator
- IBM Granite 4.1 8B model card
- Status
- active
- Retrieved at
- 2026-09-08
- Methodology
- Vendor-reported result. Vendor evaluation table; row-specific shot/CoT and task settings are retained in the card table.
GPQA (Show measurement, test conditions, and source)
- Source value
- 41.96
- Score
- 41.96
- Metric
- percent
- Unit
- %
- Category
- mixed
- Direction
- Higher is better
- Source type
- vendor-reported
- Evaluator
- IBM Granite 4.1 8B model card
- Status
- active
- Retrieved at
- 2026-09-08
- Methodology
- Vendor-reported result. Vendor evaluation table; row-specific shot/CoT and task settings are retained in the card table.
SimpleQA (Show measurement, test conditions, and source)
- Source value
- 4.82
- Score
- 4.82
- Metric
- percent
- Unit
- %
- Category
- mixed
- Direction
- Higher is better
- Source type
- vendor-reported
- Evaluator
- IBM Granite 4.1 8B model card
- Status
- active
- Retrieved at
- 2026-09-08
- Methodology
- Vendor-reported result. Vendor evaluation table; row-specific shot/CoT and task settings are retained in the card table.
AlpacaEval2 (Show measurement, test conditions, and source)
- Source value
- 50.08
- Score
- 50.08
- Metric
- percent
- Unit
- %
- Category
- mixed
- Direction
- Higher is better
- Source type
- vendor-reported
- Evaluator
- IBM Granite 4.1 8B model card
- Status
- active
- Retrieved at
- 2026-09-08
- Methodology
- Vendor-reported result. Vendor evaluation table; row-specific shot/CoT and task settings are retained in the card table.
IFEval Avg (Show measurement, test conditions, and source)
- Source value
- 87.06
- Score
- 87.06
- Metric
- percent
- Unit
- %
- Category
- mixed
- Direction
- Higher is better
- Source type
- vendor-reported
- Evaluator
- IBM Granite 4.1 8B model card
- Status
- active
- Retrieved at
- 2026-09-08
- Methodology
- Vendor-reported result. Vendor evaluation table; row-specific shot/CoT and task settings are retained in the card table.
ArenaHard (Show measurement, test conditions, and source)
- Source value
- 68.98
- Score
- 68.98
- Metric
- percent
- Unit
- %
- Category
- mixed
- Direction
- Higher is better
- Source type
- vendor-reported
- Evaluator
- IBM Granite 4.1 8B model card
- Status
- active
- Retrieved at
- 2026-09-08
- Methodology
- Vendor-reported result. Vendor evaluation table; row-specific shot/CoT and task settings are retained in the card table.
GSM8K (Show measurement, test conditions, and source)
- Source value
- 92.49
- Score
- 92.49
- Metric
- percent
- Unit
- %
- Category
- mixed
- Direction
- Higher is better
- Source type
- vendor-reported
- Evaluator
- IBM Granite 4.1 8B model card
- Status
- active
- Retrieved at
- 2026-09-08
- Methodology
- Vendor-reported result. Vendor evaluation table; row-specific shot/CoT and task settings are retained in the card table.
GSM Symbolic (Show measurement, test conditions, and source)
- Source value
- 83.7
- Score
- 83.7
- Metric
- percent
- Unit
- %
- Category
- mixed
- Direction
- Higher is better
- Source type
- vendor-reported
- Evaluator
- IBM Granite 4.1 8B model card
- Status
- active
- Retrieved at
- 2026-09-08
- Methodology
- Vendor-reported result. Vendor evaluation table; row-specific shot/CoT and task settings are retained in the card table.
Minerva (Show measurement, test conditions, and source)
- Source value
- 80.1
- Score
- 80.1
- Metric
- percent
- Unit
- %
- Category
- mixed
- Direction
- Higher is better
- Source type
- vendor-reported
- Evaluator
- IBM Granite 4.1 8B model card
- Status
- active
- Retrieved at
- 2026-09-08
- Methodology
- Vendor-reported result. Vendor evaluation table; row-specific shot/CoT and task settings are retained in the card table.
Page 1 of 4
Popularity
46.8 / 100
Sources and data date
| Type | Evidence and data date |
|---|---|
| Speed | Median across serving providers. Output speed after the first streaming chunk, not total response time. Retrieved September 8, 2026. |
| Price | Base API prices without caching or batch discounts. Higher context tiers and other rates are listed under Costs. |
| Input | Documented modalities. Retrieved September 8, 2026. |
| Output | Documented modalities. Retrieved September 8, 2026. |
| Research status | Research date September 8, 2026. 2 source URLs checked. This documents the inspected sources, not an exhaustive inventory of every publication. |
| Not found in the inspected sources | active-parameters-billions Original wording The exact card describes a dense 8B model and does not publish a separate active-parameter count. |
| Not found in the inspected sources | knowledge-cutoff Original wording The exact Granite 4.1 8B card does not publish a knowledge cutoff. |
| Additional source | OpenRouter: Popularity index (derived), CC BY 4.0 |
| Additional source | Vercel AI Gateway: Popularity index (derived), CC BY 4.0 |
| Additional source | Creative Commons Attribution 4.0 |