GPT-5.6 Terra vs o4 Mini (2025-04-16)
Humanity's Last Exam Humanity's Last Exam: Very hard expert questions across many fields, written to stump today’s models
GPT-5.6 Terra33.3
o4 Mini (2025-04-16)—
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
GPT-5.6 Terra1
o4 Mini (2025-04-16)—
AutomationBench-AA AutomationBench-AA: Agents orchestrate SaaS workflows through APIs; objectives receive credit only when the task has no guardrail violation
GPT-5.6 Terra33.5
o4 Mini (2025-04-16)—
Arena Text · creative writing Arena Text · creative writing: Human pairwise preferences in the stated text vote pool
GPT-5.6 Terra · xhigh1,429
o4 Mini (2025-04-16)—
MMMU-Pro (vision) MMMU-Pro (vision): College-level questions that require understanding images: charts, diagrams, photos
GPT-5.6 Terra76.8
o4 Mini (2025-04-16)—
| Benchmark | GPT-5.6 Terra | o4 Mini (2025-04-16) |
|---|---|---|
| Humanity's Last Exam | 33.3 | — |
| Terminal-Bench 4.0 | 1 | — |
| AutomationBench-AA | 33.5 | — |
| Arena Text · creative writing | 1,429 · xhigh | — |
| MMMU-Pro (vision) | 76.8 | — |
Specs
| Spec | GPT-5.6 Terra | o4 Mini (2025-04-16) |
|---|---|---|
| Input, per 1M | $2 | — |
| Output, per 1M | $12 | — |
| Cost per task? | $0.18 | — |
| Speed, tok/s | 125 | — |
| Context | 1.05M | — |
| Weights | Closed | Not verified |
| Released | Jul 2026 | — |
| See images | ✓ | ✓ |
| Read PDFs | Partial | ? |
| Hear audio | — | — |
| Watch video | — | — |
| Call tools | ✓ | ✓ |
| Force a tool call | Partial | ? |
| Return strict JSON | ✓ | ✓ |
| Search the web | ✓ | ? |
| Run code | ✓ | ? |
| Use a computer | ✓ | ? |
What breaks when you switch
GPT-5.6 Terra → o4 Mini (2025-04-16)
Context: not verified
o4 Mini (2025-04-16) → GPT-5.6 Terra
Prompts with >272K input tokens›
Above 272K input tokens, input costs 2× and output 1.5× for the full request.