Granite 4.1 8B vs Granite 4.2 8B
Different efforts: non-thinking · thinking
Humanity's Last Exam Humanity's Last Exam: Very hard expert questions across many fields, written to stump today’s models
Granite 4.1 8B3.8
Granite 4.2 8B9.7
Terminal-Bench 4.0 Terminal-Bench 4.0 · source setting 2: Agents use a terminal to complete executable coding, system administration and research tasks in the 4.0 suite
Granite 4.1 8B—
Granite 4.2 8B0
AutomationBench-AA AutomationBench-AA: Agents orchestrate SaaS workflows through APIs; objectives receive credit only when the task has no guardrail violation
Granite 4.1 8B—
Granite 4.2 8B2.1
Arena Text · creative writing Arena Text · creative writing: Human pairwise preferences in the stated text vote pool
Granite 4.1 8B1,263
Granite 4.2 8B1,201
| Benchmark | Granite 4.1 8B | Granite 4.2 8B |
|---|---|---|
| Humanity's Last Exam | 3.8 | 9.7 |
| Terminal-Bench 4.0 | — | 0 |
| AutomationBench-AA | — | 2.1 |
| Arena Text · creative writing | 1,263 | 1,201 |
Specs
| Spec | Granite 4.1 8B | Granite 4.2 8B |
|---|---|---|
| Input, per 1M | — | $0.06 |
| Output, per 1M | — | $0.25 |
| Cost per task? | — | $0.024 |
| Speed, tok/s | — | 65 |
| Context | 131K | 128K |
| Weights | Open | Open |
| Released | Apr 2026 | Aug 2026 |
| Memory (4-bit) | 4.4 GB | 4.4 GB |
| Licence | Apache-2.0 | Apache-2.0 |
| Commercial use | Allowed | Allowed |
| See images | — | — |
| Read PDFs | ? | — |
| Hear audio | — | — |
| Watch video | — | — |
| Call tools | ✓ | ✓ |
| Force a tool call | ? | Partial |
| Return strict JSON | Partial | ✓ |
| Search the web | ? | ? |
| Run code | ? | ? |
| Use a computer | ? | ? |
What breaks when you switch
Granite 4.1 8B → Granite 4.2 8B
Context: 131K → 128K
Granite 4.2 8B → Granite 4.1 8B
Nothing known.