Exact version
Check the model identifier, date and whether the alias can change.
DOCUMENTED BENCHMARK
Agents' Last Exam
Published results and conditions declared by their sources. Figures are comparable only when version, configuration, metric and date match.
01 / RESULTS
Published measurements
| Model | Organization | Result | Metric | Conditions |
|---|---|---|---|---|
| GPT‑5.6 Sol | OpenAI | 52,7 % | Accuracy | Official GPT‑5.6 table. |
| GPT‑5.6 Terra | OpenAI | 50,4 % | Accuracy | Official GPT‑5.6 table. |
| GPT‑5.6 Luna | OpenAI | 50,3 % | Accuracy | Official GPT‑5.6 table. |
02 / INTERPRETATION
Before comparing
Equivalent configuration
Review tools, effort, number of attempts, prompt and inference budget.
Transfer to your use case
Validate the result with tasks, languages, formats and errors representative of your product.
03 / METHOD
From benchmark to reproducible decision
A public benchmark provides guidance, but does not replace an evaluation with your data, tools, budget, and error tolerance.
Identify the exact version
A moving alias and a snapshot may produce different results. Record the model, date, and provider.
Reconstruct the conditions
Prompt, reasoning, tools, number of attempts, and budget must be equivalent for comparison.
Look for uncertainty
A small difference may disappear between runs. Keep samples, dispersion, and failures, not just the average.
Check transferability
Validate whether the improvement holds across languages, formats, and real cases close to your product.
A metric is only useful when we know the task, the conditions, and the error that matters. The result begins with the design of the test.
ANALYSIS
Related analysis
AutomationBench: What an Agent That Leaves a Business Workflow in the Right State Demonstrates—and What It Does Not
AutomationBench evaluates whether an agent can complete workflows across simulated SaaS applications and bring them to a verifiable final state. This guide explains what that evidence means, how to read its metrics, and what additional testing is needed before extrapolating a result to a real company.
22 Sep 2026 ↗
Agents’ Last Exam: what it measures about a real-work agent—and why its pass rate does not mean “automating a job”
Agents’ Last Exam evaluates agents on professional tasks within operating-system environments and with verifiable outcomes. Its design is useful for studying workflow execution, but an aggregate success figure does not demonstrate that a job can be automated. Interpreting it requires details about the tasks, environment, harness, model, tools, budget, and evaluated version.
22 Sep 2026 ↗04 / SOURCES