Pick one list, not a number: about 75 problems if you have one to three months, about 150 if you have longer. Count a problem as done only after a clean re-solve days later, and stop adding new problems when unseen mediums from your list's patterns go well in mixed, timed sets.
The honest answer: coverage and re-solves, not a count
We found no employer page that states a required number of solved problems, and no study that links a solved count to offers. What exists is guidance from the people who publish the lists, and it clusters in a narrow range. LeetCode labels its LeetCode 75 plan as best for one to three months of preparation and its Top Interview 150 plan as best for three months or more. The author of Grind 75, a former Meta staff engineer, writes in its FAQ that the returns on doing questions beyond its 169-question pool are diminishing. None of these pages tells you to work through thousands.
A count is a poor target because of what the interview measures. You will be asked to solve a problem you have not seen while explaining your reasoning, which depends on two things: whether you have met the underlying pattern, and whether you can still execute it. Three hundred problems solved once and forgotten leave you with neither. Eighty problems across the main patterns, each of which you can rebuild from a blank file, give you both. The rest of this guide turns that into numbers you can plan with.
How many LeetCode problems are there?
As of November 14, 2026, LeetCode's problem list shows 4,110 problems, and the public data behind the page splits them into 975 Easy, 2,146 Medium and 989 Hard. The total spans every category tab; the same data puts 3,692 of the problems in Algorithms, with the rest under Database, Shell, Concurrency, JavaScript and pandas. On the same day the topic filter showed 2,281 problems tagged Array, 901 String, 847 Hash Table and 689 Dynamic Programming; a problem can carry several tags, so those counts overlap. All of these figures change as problems are added, so check the live page for the current ones.
The number is useful mainly as a reason not to chase it. At about nine new problems a week, the pace an eight-hour weekly budget supports in the table further down, the full catalog would take more than eight years. The lists in the next section exist because a small fraction of the catalog covers the patterns that interviews draw on.
| Difficulty | Problems | Share |
|---|---|---|
| Easy | 975 | 23.7% |
| Medium | 2,146 | 52.2% |
| Hard | 989 | 24.1% |
| All | 4,110 | 100% |
How big the curated lists are: Blind 75, Grind 75, NeetCode 150, LeetCode 75
The Blind 75 is the list most others descend from. Its author, who also wrote the Tech Interview Handbook, says he curated 75 questions and shared them on Blind, that someone reposted the list on the LeetCode forum, and that the community gave it its name. His newer Grind 75 tool draws on a pool of 169 questions and, with default settings, schedules 75 of them over 8 weeks at 8 hours a week; you can change the weeks and hours to fit your own schedule. NeetCode describes its 150 as the Blind 75 plus 75 more problems and its 250 as the 150 plus 100 more, aimed at people new to algorithms. LeetCode publishes two official plans of its own, LeetCode 75 and Top Interview 150.
Every figure in the table was read from the publisher's own page on November 14, 2026, except the Blind 75 difficulty mix, which comes from NeetCode's copy of the list; lists are revised, so confirm before you plan around one. Two things are worth noticing. The lists are mostly Medium, and some contain others by construction: NeetCode 150 includes the Blind 75, and NeetCode 250 includes the 150. Stacking lists adds less than it appears to, so pick one and finish it.
| List | Problems | Difficulty mix | Publisher's note |
|---|---|---|---|
| Blind 75 | 75 | 19 / 49 / 7 (as listed on neetcode.io) | First shared on Blind by the Tech Interview Handbook author |
| Grind 75 | 75 by default; pool of 169 | 24 / 42 / 9 | Default plan: 8 weeks at 8 hours a week |
| NeetCode 150 | 150 | 28 / 101 / 21 | Blind 75 plus 75 more |
| NeetCode 250 | 250 | 60 / 155 / 35 | NeetCode 150 plus 100 more; for beginners |
| LeetCode 75 | 75 | 22 / 53 / 0 | LeetCode: best for 1 to 3 months of prep |
| Top Interview 150 | 150 | 39 / 93 / 18 | LeetCode: best for 3+ months of prep |
How many LeetCode problems per day? A pace by weekly time budget
Grind 75 offers a rare published time estimate. With default settings its page puts the time needed for the 75 questions at 64 hours, about 51 minutes each. The recommended solving times on the same page run from 15 to 45 minutes per question and add up to 32.25 hours; the page explains that its estimate assumes roughly as much time again to check the answer to each question. The table below is our arithmetic on that 51-minute figure: a rule of thumb, not a measurement. A 150-problem list with more Hard problems will run slower, and re-solves of missed problems come on top. The Tech Interview Handbook's separate study plan recommends three months at 11 hours a week for a fuller preparation.
Per day, eight hours a week works out to one or two problems, but a daily quota is the wrong unit. A weekly block that includes re-solves holds up better than a streak of one rushed problem a night. Expect to be slower than the table at the start. The Grind 75 FAQ suggests a checkpoint: if you still need longer than the recommended duration after 20 to 30 questions, it calls that a sign that you need more revision of fundamentals. And do not buy speed by skipping the answer-checking half of the estimate; it is the half that makes the first half stick.
| Hours per week | New problems per week | Weeks for 75 | Weeks for 150 |
|---|---|---|---|
| 5 | About 6 | 13 | 26 |
| 8 | About 9 | 8 | 16 |
| 11 | About 13 | 6 | 12 |
| 15 | About 18 | 5 | 9 |
| 20 | About 23 | 4 | 7 |
Why re-solving counts more than new problems
The solved counter rewards the wrong event. In Roediger and Karpicke's 2006 experiment, students who studied a passage in four periods recalled 40 percent of it a week later, while students who studied it once and practiced recalling it three times recalled 61 percent. The study used prose, not code, but the lesson fits: a problem you read the solution to, or solved once with a hint, is in the rereading column. A problem you rebuilt from a blank file several days later is in the recall column.
So change the definition of done. Use the four statuses below and count only the last two toward your target. This usually shrinks the number you can honestly claim and improves how you perform on an unseen question. The companion guide on remembering solutions has a re-solve schedule and a tracker, and a spaced-repetition app such as RecallDeck can keep the concept layer (complexity, data structure behavior, language details) in rotation while you re-code.
| Status | Definition | Counts toward the target? |
|---|---|---|
| Attempted | Got stuck and read the solution | No |
| Solved once | Accepted, perhaps with a hint or over time | Not yet: schedule a re-solve |
| Re-solved | Clean, from a blank file, three or more days later | Yes |
| Transferred | Solved an unseen problem of the same pattern | Yes; the pattern is covered |
Stop criteria: when you have done enough
Readiness shows up as behavior on problems you have not seen, not as a total. The criteria below are our rules of thumb, not validated thresholds. Run this check every two weeks: take four unseen Medium problems from the topics on your list, mix them, and give each the time you will have in the real round. If the criteria below hold, more new problems will add little, and your hours are better spent on re-solves, mock interviews and the other rounds. If they do not hold, the misses tell you which pattern to return to.
- You finished one curated list, and every problem you missed has a clean re-solve at least three days after the miss.
- In the mixed, timed check you solve about three of four unseen Mediums and can state the complexity of each.
- You name the pattern and its invariant in the first few minutes, before writing code.
- Your misses come from a new twist, not from a forgotten template or a familiar bug.
- System design, behavioral stories or role fundamentals are now the weaker part of your loop.
How many for FAANG, for a first job, or with two weeks left
For large companies the honest answer does not change: none of the sources above publishes a required count, and the Grind 75 pool, built by a former Meta staff engineer, stops at 169 problems. LeetCode's study plan page does list short company sets, such as 25 recently asked Amazon questions and 23 for Google, labeled Spring '23, which suit the last week or two after a general list. Priorities inside a list also differ by target: the Tech Interview Handbook's author calls dynamic programming questions less important, to be done if you have time to spare, while noting that companies like Google still ask them.
Otherwise let the calendar decide. The time figures below come from the pace table, so adjust them to your own hours. If you are returning to interviews after a few years, start by re-solving problems you already did; the misses will show you which part of a list you need.
- Two weeks left: about 20 to 35 problems across the highest-priority topics, with re-solves; do not start a 150 list.
- One to three months: one 75-problem list, finished and re-solved.
- Three months or more, or shaky fundamentals: a 150-problem list; NeetCode describes its 250 as its most beginner-friendly list.
- Any timeline: stop new problems a few days before the interview and re-solve your misses.
Quick answers
Frequently asked questions
How many LeetCode problems are there?
LeetCode's problem list showed 4,110 problems on November 14, 2026; the public data behind the page splits them into 975 Easy, 2,146 Medium and 989 Hard. The figure changes over time, so check the problem list for the current total.
How many LeetCode problems should I do per day?
One or two a day is what an eight-hour week supports at about 51 minutes per problem with answer checking, the estimate behind Grind 75's default plan. Plan by the week instead of the day, and include re-solves of earlier misses in that time.
How many LeetCode problems do I need for FAANG?
We could not find a company that publishes a required number. Grind 75, curated by a former Meta staff engineer, draws on a pool of 169 problems, and its FAQ describes diminishing returns beyond the list. Coverage of patterns and the ability to re-solve matter more than the total.
Is 100 LeetCode problems enough?
It can be, if the hundred cover the main patterns and you can re-solve the ones you missed from a blank file. A hundred problems solved once and forgotten is not enough, and neither is three hundred. Test yourself with unseen Medium problems under time pressure.
Should I do Easy, Medium or Hard problems?
Mostly Medium. The curated lists lean that way: NeetCode 150 has 101 Medium problems out of 150, and LeetCode 75 has 53 Medium, 22 Easy and no Hard. Use Easy problems to learn a pattern and add Hard ones after the Mediums feel routine.
Source notes
References and review policy
Information checked on November 14, 2026. Section links identify sources for factual claims and technical explanations. Interpretations, practice scenarios and preparation recommendations are RecallDeck’s editorial work.
- LeetCode: Problems
- LeetCode 75 study plan
- LeetCode Top Interview 150 study plan
- LeetCode: Study Plan catalog
- Tech Interview Handbook: Best practice questions (Blind 75)
- Grind 75 (Tech Interview Handbook)
- Grind 75: FAQ
- Tech Interview Handbook: Coding interview study plan
- NeetCode 150
- NeetCode 250
- NeetCode: Blind 75
- Roediger & Karpicke (2006), full text (Purdue Learning Lab copy)
From reading to recall
Practice the full interview loop.
RecallDeck schedules the concepts you miss and keeps coding, design, and behavioral fundamentals available when the interviewer changes direction.