GPT-4.1 Nano vs GPT-6 Luna
Different efforts: non-thinking · medium
Humanity's Last Exam Humanity's Last Exam: Very hard expert questions across many fields, written to stump today’s models
GPT-4.1 Nano3.8
GPT-6 Luna28.3
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-4.1 Nano—
GPT-6 Luna2.5
AutomationBench-AA AutomationBench-AA: Agents orchestrate SaaS workflows through APIs; objectives receive credit only when the task has no guardrail violation
GPT-4.1 Nano—
GPT-6 Luna40.5
Arena Text · creative writing Arena Text · creative writing: Human pairwise preferences in the stated text vote pool
GPT-4.1 Nano1,307
GPT-6 Luna · max1,410
MMMU-Pro (vision) MMMU-Pro (vision): College-level questions that require understanding images: charts, diagrams, photos
GPT-4.1 Nano40.1
GPT-6 Luna77.9
| Benchmark | GPT-4.1 Nano | GPT-6 Luna |
|---|---|---|
| Humanity's Last Exam | 3.8 | 28.3 |
| Terminal-Bench 4.0 | — | 2.5 |
| AutomationBench-AA | — | 40.5 |
| Arena Text · creative writing | 1,307 | 1,410 · max |
| MMMU-Pro (vision) | 40.1 | 77.9 |
Specs
| Spec | GPT-4.1 Nano | GPT-6 Luna |
|---|---|---|
| Input, per 1M | $0.10 | $0.10 |
| Output, per 1M | $0.40 | $0.50 |
| Cost per task? | — | $0.017 |
| Speed, tok/s | 263 | — |
| Context | 1M | 1.05M |
| Weights | Closed | Closed |
| Released | Apr 2025 | Sep 2026 |
| See images | ✓ | ✓ |
| Read PDFs | Partial | Partial |
| Hear audio | — | — |
| Watch video | — | — |
| Call tools | ✓ | ✓ |
| Force a tool call | Partial | Partial |
| Return strict JSON | ✓ | ✓ |
| Search the web | ? | ✓ |
| Run code | ? | ✓ |
| Use a computer | ? | ✓ |
What breaks when you switch
GPT-4.1 Nano → GPT-6 Luna
Function calling only with none · Chat Completions›
Chat Completions function calling requires reasoning_effort none; use Responses for reasoning with tools.
More than 272K input tokens›
Prompts above 272K input tokens charge 2× input and cache rates and 1.5× output rates for the entire request.
Remove unsupported sampling parameters›
When reasoning effort is not none, remove temperature, top_p and top_logprobs, plus Chat Completions logprobs or Responses output logprobs inclusion.
Cache writes billed at 1.25x›
Cache writes cost 1.25× the uncached input token rate.
GPT-6 Luna → GPT-4.1 Nano