Quick wins for a faster PC:
Repair Windows errors before they cause bigger problemsFix Now →Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Some links on this page are affiliate links: if you buy through them we may earn a commission, at no extra cost to you.
“Strawberry” contains three lowercase r’s—at positions 3, 8 and 9:
s – t – r – a – w – b – e – r – r – y
Some language models have historically answered “two” because counting letters is not the same task as recognizing, spelling or discussing a word. Large language models primarily process tokens and generate likely continuations; they are not automatically running a guaranteed, letter-by-letter counting routine.
The strawberry test is a narrow but revealing failure
The question became widely discussed in 2024 after people found that some language models could write sophisticated explanations, code and mathematics yet miscount the r’s in “strawberry.” The example is memorable because the answer appears obvious to a person who simply inspects the word.
#1 Best Overall
- Package Contents: You will receive a set of 5 pack A5 Kraft notebooks, each containing 30 sheets (60 pages). These notebooks are highly cost-effective, affordable, and made with premium quality materials, making them an excellent choice for school or everyday use
- Study Paper: These work notebooks are designed for easy note-taking, featuring a 180° lay-flat design and secure binding to prevent pages from falling out. The durable Kraft paper cover ensures long-lasting use, while the beige inner pages help reduce eye strain and provide a comfortable writing experience. The thick, high-quality paper prevents ink bleed-through, making it ideal for writing or sketching
- Suitable Size: With dimensions of 8.15 x 5.5 inches, these subject notebooks are compact and portable. They fit perfectly in handbags, backpacks, or even clothing pockets, making them convenient to carry wherever you go
- Efficient Office & Study: Our A5 lined notebooks feature ruled inner pages, making them perfect for middle school, high school, and college students to take notes, complete homework, or organize subjects. The ruled lines help keep handwriting neat and tidy. These notebooks are also ideal for office use, whether for taking meeting minutes, planning daily tasks, or jotting down important ideas. The paper are made of FSC-Certified wood
- Wide Range of Uses: These office supplies notebooks are perfect for offices, classrooms, meetings, or business settings. They’re great for homework, journaling, sketching, planning, or jotting down important notes. They also make thoughtful and practical gifts for friends, family, or colleagues
It is important not to overstate the result. Not every AI system gets this question wrong, and newer models often answer the original question correctly. Results can vary with the model version, prompt, sampling settings, system instructions, reasoning mode and access to tools. The strawberry question is now better understood as a historical diagnostic and an entry point into a broader limitation: language models can be unreliable at exact character-level operations, especially on unusual or adversarial inputs.
What an AI model processes: tokens, not necessarily letters
Before text enters a large language model, it is converted into tokens. A token can be a complete common word, a word fragment, punctuation, a space-plus-word sequence or a byte-level sequence, depending on the model’s tokenizer and vocabulary.
For illustration, a tokenizer might represent a familiar word using chunks resembling straw and berry. But that split is not universal: different models can divide the same word differently, and a word may be represented as one token or several.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
This matters because the model’s basic computational units are not necessarily individual characters. A human solving the problem can deliberately scan the word like this:
- Inspect the next character.
- Compare it with
r. - Increase a running count when it matches.
- Continue until the word ends.
A language model’s ordinary operation is different: it processes token IDs and their context, then predicts which token is likely to come next. That representation can contain information about spelling, but it does not automatically impose a reliable loop over every character.
Research has specifically connected tokenization with variation in large language models’ counting performance. A 2024 study on counting ability and tokenization examined how the way text is divided into tokens can affect such tasks.
Rank #2
- Perfect size: 19cm x 13cm/ 7.5 "x 5.1", perfect size for handbag, schoolbag or backpack, easy Blank take pages for running.
- Features: 50 sheets (100 pages) of blank pages per book. Perfect for sketching and notes. Portable size.
- Material: Strong brown hard cover and blank cream white paper, thick paper prevents ink from inks through the pages, and the binding of each spiral notebook keeps these pages together.
- Wide usage: Ideal for a diary, travel journal, poetry work, creativ e writing, making sketches and drawings, Work records, study notes, mood diary, scrapbooks and so on.
Tokenization is important—but it is not the whole explanation
It would be wrong to say that tokens make letters invisible. A model may learn the internal character structure of a token, and it can often reproduce the spelling of a familiar word accurately. The more precise point is that character information is not always directly exposed in the model’s earliest representation or reliably retrieved for an exact query.
Free tools Windows power users keep installed
One-click scans. No signup required.
Research published in 2025 found that language models could spell tokens character by character with high accuracy while still struggling with more complex tasks involving the internal composition of tokens. The work suggests that character-level information can be reconstructed in later Transformer layers rather than being fully available at the initial embedding stage.
So the model is not literally blind to the letters in “strawberry.” It may have learned them. The problem is whether it will access and manipulate that information in the precise way the question requires.
Spelling, counting and understanding are different capabilities
A model can know that a strawberry is a fruit, produce a fluent description of one and reproduce the correctly spelled word without reliably counting its letters.
Spelling a familiar word can be a strong pattern-completion task. The sequence strawberry is highly familiar, so generating it as a whole may be easy. Counting requires a different procedure:
Recommended Free Tools
- isolating each occurrence of a target character;
- maintaining an exact running total;
- distinguishing written form from pronunciation and meaning; and
- checking the result instead of stopping at a plausible answer.
This is why a model might spell the word correctly and then give the wrong count. Correct reproduction does not prove that the model performed a character-by-character inspection.
Rank #3
- Sturdy Construction: Our Lined Spiral Journal Notebook is built to last with a sturdy metal twin-wire binding and a tough hardcover. The water-resistant cover shields your notes from damage, while the double-wire design allows for easy folding and flat laying.
- High-Quality Paper: Crafted from 100 GSM thick, ink-friendly paper, our notebook prevents ink bleed-through and ghosting. It accommodates various pens, including ballpoint, gel, and fountain pens. Each page features a day header for effortless date tracking.
- Organized and Functional Design: With 140 lined pages and a 6-page blank table of contents, our notebook offers ample space for note-taking and easy referencing. An inner pocket keeps miscellaneous items secure, and an elastic closure band ensures the notebook stays closed when not in use.
- Versatile Usage: Suitable for office, school, and home environments, our notebook is perfect for journaling, note-taking, drawing, goal setting, Bible, and planning. It's a thoughtful present for friends, family, classmates, and colleagues.
- Medium-Sized Portability: Measuring 5.7 inches x 7.9 inches, our medium notebook strikes the perfect balance between portability and functionality. Its sturdy construction and aesthetic design make it an ideal companion for all your writing endeavors.
Why a confident answer can still be wrong
Calling every such mistake a “hallucination” is imprecise. It is certainly a factual error, and it may fit the broad definition of an unsupported generated claim. But the more useful description is a failure to execute or verify an exact symbolic operation.
Language models generate outputs probabilistically. They perform complex learned computation, but their normal output mechanism does not guarantee the result in the way a string-counting function does. A fluent answer can therefore reflect a likely continuation rather than a verified calculation.
The model may recognize the word semantically, associate it with common language patterns and produce a number that sounds plausible in context. That does not establish that it has run a deterministic count. It also does not prove that a particular internet misspelling or false example caused the error; that explanation remains speculative without evidence about the model’s training and computation.
Why asking the model to spell the word first can help
A more structured prompt changes the task:
Write “strawberry” one letter at a time, then count the
r’s.
This encourages an intermediate representation:
s, t, r, a, w, b, e, r, r, y
The count is then easier to inspect. However, “show your work” is not a guarantee. The model could generate an incorrect intermediate spelling or miscount the displayed sequence. Treat the sequence as something to verify, not as proof that the model’s hidden computation followed that exact path.
Why reasoning models may perform better
More deliberate reasoning can give a model additional opportunities to decompose the task, spell out the word, check an initial answer or try another strategy. That helps explain why reasoning-capable models can outperform ordinary chat models on some character-level questions.
Rank #4
- Scan, study and organize your notes with the Five Star Study App. Create instant flashcards and sync your notes to Google Drive to access them anywhere from any device.
- This 3 subject notebook has 150 double-sided, college ruled sheets that fight ink bleed and are perforated for easy tear out. Sheets measure 8-1/2" x 11" when torn out.
- Tough pockets help prevent tears and hold 8-1/2" x 11" loose sheets. Durable plastic front cover is water-resistant to help protect your notes and our Spiral Lock wire helps prevent snags on clothes and backpacks.
- Made with SFI certified paper. Notebook is recyclable – just remove the reinforcement tape on the pocket and recycle the rest! Available in Blue (Color May Vary)
- LASTS ALL YEAR. GUARANTEED!*
OpenAI’s September 12, 2024 article, “Learning to reason with LLMs,” described o1-preview as a model trained with reinforcement learning to spend more time reasoning, recognize mistakes and try alternative approaches. The public demonstration included a character-decoding task whose answer stated that there are three r’s in “strawberry.” OpenAI also described additional train-time and test-time computation as part of the model’s approach.
That demonstration should be interpreted carefully. It shows that additional reasoning can help with a particular task; it does not prove universal, human-like character awareness. Reasoning models can still fail on long strings, rare words, misspellings, positions, Unicode variants and deliberately confusing inputs.
Where similar weaknesses appear
The same general mismatch can surface in questions such as:
- How many
e’s are in “experience”? - What is the seventh character of this word?
- Are these two long strings exactly identical?
- Which character differs between these strings?
- How many opening parentheses appear in this expression?
- Reverse this long string without changing anything.
- Which words in this paragraph contain exactly two
t’s?
These are related, but not identical, failure modes. Performance depends on the model and the particular input. Errors become more likely when the string is rare, nonsensical, long, intentionally misspelled or contains unusual formatting.
Useful edge cases include Strawberry, STRAWBERRY, strawberrry, stawberry and straw-berry. Unicode characters, visually similar characters, inserted punctuation, spaces and zero-width characters can introduce additional complications. A question about lowercase r is not necessarily the same as a question about every case variant or every visually similar character.
Use deterministic tools for exact string operations
For an exact count, an ordinary programming-language function is more reliable, auditable and inexpensive than asking an LLM to calculate unaided.
Best Value
- BEST-SELLING HARDCOVER JOURNAL: This classic 5.6" x 8" vegan leather journal features a durable and water-resistant cover, 160 college ruled lined pages, inner expandable pocket, sticker labels, ribbon bookmark & elastic closure band.
- PREMIUM PAPER: Made with high-quality, 100 gsm acid-free paper in light ivory color, our journal paper is thicker than average notebooks & note pads, so you can confidently use most pens, pencils, and markers without ghosting and bleed-through.
- LAY FLAT DESIGN FOR WRITING EASE: Our thread-bound, college ruled notebook is designed to lay flat, making it easier to write for both right and left-handed users. It’s the perfect notebook for journaling, note taking and planning.
- INNER POCKET: Includes an expandable inner storage pocket to store appointment cards, notes, receipts, and more. Personalize your journal cover & spine with the sheet of sticker labels included.
- VERSATILE LINED NOTEBOOK: Ideal for journaling, note-taking, planning, or creative writing. Whether you're making a to-do list, capturing ideas, or writing notes, this journal makes a perfect notebook for school, work, or home office.
Python
word = "strawberry"
count = word.count("r")
print(count) # 3See the official Python site for the language and its standard library.
JavaScript
const word = "strawberry";
const count = [...word].filter(character => character === "r").length;
console.log(count); // 3The spread operation makes the character sequence explicit for ordinary strings. Unicode grapheme handling can require more specialized logic, because a user-perceived character is not always represented by one simple code point. The MDN JavaScript documentation is the appropriate reference for language behavior.
Shell
python -c 'print("strawberry".count("r"))'For production systems, use standard string functions for counting, indexing, comparison and transformation; spellcheckers or dictionaries for spelling validation; tokenizer libraries for token-boundary questions; structured parsers for syntax-sensitive input; and calculators or symbolic systems for exact mathematics.
A practical architecture is simple: let the LLM interpret a natural-language request, pass the relevant string to a deterministic function, return the computed result, and optionally ask the model to explain it. For the strawberry question, buying a more expensive AI model solely to count letters is unnecessary.
What the strawberry example does—and does not—prove
The example does not prove that AI is unintelligent, that language models have no understanding of words or that tokenization alone explains every mistake. A system can be strong at translation, summarization, semantic analogy, factual retrieval or code generation while being weak at exact character counts, copying, bracket matching or arithmetic with carries.
It does show that AI capability is uneven and representation-dependent. Fluency, semantic knowledge, reasoning performance and exact symbolic manipulation should be evaluated separately.
The best general rule is straightforward: when a task is exact, repetitive, symbolic or easy to formalize, use a deterministic tool for the answer and use the language model for interpretation, explanation or interface design.
Windows Errors? Fix Them Before They Spread
Repair common Windows errors and clear accumulated junk for a smoother, more stable PC - no reinstall needed.Free scan · no reinstallOutdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchQuick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
