Karpathy Wiki
Where it ranks
How Karpathy Wiki compares with every other measured tool.
Best ranks
Worst ranks
Agentic Memory Indexi
MethodologyTap a logo for the tool page.
Agentic Memory Index v0.2 · Sep 2026
Check your own agent memory
See what your agent memory gets right, what it forgets, and where it fails. Tell us which memory system you build or use and we’ll show you what the overall score doesn’t tell you.
Accuracy
How often Karpathy Wiki finds the right answer, and how long the facts it stores survive.
Agentic Memory Index
v0.2Correct answers by task typei
Pass rate per task type against the field average.
Long-term retentioni
How well early facts survive: probed at five checkpoints across 56 working sessions.
Failures
How Karpathy Wiki fails when it does, and which task types trip it.
Failure Attributioni
4 failures among 150 graded cases, by class.
Failures by task typei
Failures per task type and class.
| Task type · cases | Question Not Addressed | Incorrect Answer | Outdated Answer | Fabricated Memory | Leaked Deleted Data | Not Erased | Failed |
|---|---|---|---|---|---|---|---|
| Direct Recall · 13 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
| Synthesis · 13 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
| Updated Facts · 13 | 0 | 0 | 1 | 0 | 0 | 0 | 1 |
| Time Relationships · 13 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
| False Memory Check · 10 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
| Preferences · 10 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
| Conflicting Facts · 12 | 1 | 1 | 0 | 0 | 0 | 0 | 2 |
| Erasure · 16 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
| Long Term · 50 | 1 | 0 | 0 | 0 | 0 | 0 | 1 |
Speed
How long Karpathy Wiki takes to set up, get ready, store, and answer.
Answer speedi
How long an answer takes end to end, across 29 sessions.
Speed
Cost and pricing
What Karpathy Wiki costs to run per 1,000 questions, and what it charges.
Storage and answering costi
Cost / 1,000 questions, by session kind.