African greys
The strongest evidence-backed choice for large human-mimic repertoires — which is a statement about averages across many birds, not about the one in front of you.
Records are documented attempts under one organisation's rules — not a census of every parrot alive, and not a measure of understanding. Read that way, they are genuinely useful. Read the other way, they produce most of the folklore in aviculture.
Every entry below reports a repertoire, not a comprehension test. Where a claim's paper trail is incomplete, that is stated rather than smoothed over.
Guinness World Records identifies Puck, a male budgerigar, as the bird with the largest vocabulary ever recorded: an estimated 1,728 words before his death in 1994 in Petaluma, California. It is striking partly because Puck was a budgerigar, Melopsittacus undulatus — not an African grey.
The record concerns the size of the spoken repertoire. It is not evidence that Puck understood the semantic meaning of all 1,728 items, and it should never be presented that way.
Guinness's live online record names Oskar, owned by Gabriela Danisch in Germany, with 148 words, verified 8 September 2010.
Guinness itself warns that records change daily and its web pages are not necessarily updated immediately. So the safe wording is “the current Guinness-published record for largest vocabulary in a living bird.” Do not write “no living bird knows more than 148 words.” A Guinness record is a documented attempt under that organisation's rules — not a comprehensive scientific census of every companion parrot alive.
Prudle, an African grey, is repeatedly described in historical references as an earlier Guinness vocabulary record holder with roughly 800 words. The claim is plausible and very widely reproduced, but our research pass did not locate a current Guinness primary record page preserving the original documentation.
It is therefore listed here as a historical Guinness-era claim requiring archival confirmation — not as equal in verification status to Puck's current official page.
Sparkie, a British budgerigar born in 1954, was a celebrated competition talker who won the BBC International Cage Word Contest in 1958. Historical accounts attribute 531 words, 357 phrases, and eight nursery rhymes to him, and surviving BBC archival material and museum accounts support his extraordinary talking reputation.
Those numerical counts are historical rather than the product of a modern experimental vocabulary protocol. They carry a historical-record label here, not a “scientifically tested comprehension” one.
Irene Pepperberg's decades of work with the African grey Alex used controlled transfer tests to determine whether learned vocal labels were functionally related to objects, colours, shapes, materials, relations, and quantities. Alex performed on categories such as same/different, could label properties, and answered numerical questions under experimental conditions. The work was about symbolic, reference-like behaviour — not imitation.
In numerical experiments Alex demonstrated cardinal quantity competence and generalised the vocal label “none” to tasks where no member of a requested category was present. Pepperberg and Gordon described this as a “zero-like concept” — an appropriately careful phrase. It was evidence that Alex could use an absence label in a numerically relevant way, not proof that his abstract concept of zero matched an educated human's.
These results establish that at least one intensively trained African grey could solve those problems. They do not establish that every grey spontaneously understands numbers or categories. Griffin and other greys later provided additional data in related paradigms, strengthening the case that Alex was not biologically unique — but the experimental literature remains tiny compared with human developmental psychology.
Pepperberg & Gordon, Journal of Comparative Psychology 119, 197–209, 2005, 10.1037/0735-7036.119.2.197Apollo, a companion African grey, was recognised by Guinness after identifying 12 items in three minutes, achieved in 2023. That is a genuine controlled record achievement.
Apollo's owners and Guinness also publicise broader claims about colours, counting, question answering, and “toddler-level” intelligence. Unless a particular capability has been studied under an independent peer-reviewed protocol, this site classifies it as viral or owner-demonstrated, not scientifically established — and keeps it on the famous talkers page.
Calling any parrot “as intelligent as a human toddler” is especially misleading. Cognitive ability is multidimensional: a parrot may equal or outperform young children on a particular perceptual, memory, statistical, or categorisation task while lacking many other human competencies. A “human toddler” headline is not a standardised comparative-psychology diagnosis.
Species differed strongly in owner-reported mimicry repertoire, with African greys showing the largest repertoires overall. Across the full dataset there was no universal male advantage, though sex differences occurred in five of the 19 best-sampled species — budgerigars among the species where males showed larger mimicry repertoires.
That budgie sex difference fits earlier experimental work in which males often modify calls to converge on female partners during pair formation, while the reverse imitation is less pronounced in that context.
Benedict et al., Scientific Reports 12, 20271, 2022, 10.1038/s41598-022-24335-xThe strongest evidence-backed choice for large human-mimic repertoires — which is a statement about averages across many birds, not about the one in front of you.
Can be extraordinary despite their size, particularly males. Both all-time vocabulary records on this page are budgies. Individual variation is still enormous.
Widely successful human-speech mimics, though the comparative scientific literature is thinner than avicultural lore implies.
Absolutely capable of learned human sounds and words — and should not be marketed as predictably high-vocabulary species.
Cookie, a Major Mitchell's cockatoo at Brookfield Zoo, was at least 82 years and 88 days old when he died on 27 August 2016. Guinness bases that minimum on zoo documentation: Cookie arrived in May 1934 and was estimated to be at least a year old. The zoo assigned a hatch date of 30 June 1933, which would imply a greater age still — but Guinness conservatively records the documented minimum.
It is a perfect illustration of why a records page should distinguish a verified minimum age from a retrospectively estimated birthday.
Guinness recognises Spoodles, a plum-headed parakeet, at 32 years and 86 days, verified in Charleston, South Carolina as of 28 March 2026, with a Guinness feature published on 26 June 2026.
The gap between 32-year-old Spoodles and 82-year-old Cookie illustrates the essential record-keeping issue: “oldest living Guinness-certified parrot” is not the same as “oldest parrot alive somewhere today.” Guinness requires documentation and a successful claim, and notes that record status can change faster than its pages update.
A major 2022 comparative study assembled records for more than 130,000 individual parrots from over 1,000 zoological institutions, covering 217 species. Estimated average captive life expectancy ranged enormously — from about two years in some fig parrots to about 30 years in scarlet macaws, with sulphur-crested cockatoos around 25 years in that dataset. Some exceptional individuals exceeded 80 years.
This retires one of aviculture's most persistent errors. Cookie surviving past 82 does not mean a newly acquired cockatoo has an expected lifespan of 80 years. The same analysis also found an evolutionary relationship between relative brain size and life expectancy among parrots — a comparative correlation supporting hypotheses about behavioural flexibility and slow life histories, not a claim that smarter individual parrots live longer because they solve more puzzles.
Smeele et al., Proc. R. Soc. B 2022Reliable individual ages are far easier to obtain from managed populations than from long-lived wild parrots, where an animal may simply disappear from observation — died, dispersed, or evaded detection, with no way to tell which. It is therefore not defensible to publish a universal table claiming, for example, “African greys live exactly 23 years wild and 60 years captive.” The strongest large-scale dataset concerns managed populations.
Olkowicz and colleagues showed that parrots and songbirds pack exceptionally high densities of neurons into the forebrain, so some bird species possess forebrain neuron numbers comparable with or exceeding those of much larger-brained primates. The paper did not say a parrot brain is a primate brain — avian and mammalian brains differ substantially in development and gross organisation.
Relative brain size was historically a useful comparative measure, but it cannot rank cognition cleanly. Absolute neuron number, distribution (particularly pallial neuron number), connectivity, circuit specialisation, ecology, developmental history, and the specific task all matter. A “brain size divided by body size” leaderboard is biologically inadequate — which is covered further on the cognition page.
Olkowicz et al., PNAS 113, 7255–7260, 2016, 10.1073/pnas.1517131113“Smartest bird” has no answer. Almost every other superlative does — which is more interesting, not less.
| Question | Holder | Evidence |
|---|---|---|
| Best documented human-speech vocabulary | Puck, budgerigar — 1,728 words | GWR verified |
| Largest vocabulary in a living bird, as currently published | Oskar, budgerigar — 148 words | GWR verified |
| Most scientifically studied symbolic talking parrot | Alex, African grey | Proven |
| Classic spontaneous tool-manufacture parrot | Goffin's cockatoo — Figaro, then 12 more | Proven |
| Landmark probabilistic-reasoning result | Kea — six birds | Addendum |
| Famous beat-synchronising bird | Snowball, sulphur-crested cockatoo | Proven |
| Oldest Guinness-certified parrot ever | Cookie, Major Mitchell's cockatoo — 82 yr 88 d | GWR verified |
| Currently published oldest living parrot | Spoodles, plum-headed parakeet — 32 yr 86 d | GWR verified |
| Highest intelligence | No scientifically valid universal record exists | Not a record |
False. Puck and the currently published living-vocabulary record are both budgies.
False at the order level. The 877-bird survey found no overall sex effect; sex differences are species-specific.
Confuses vocal production with semantic competence. They are separate measurements requiring separate tests.
Confuses exceptional maxima with average survival. See the 130,000-bird dataset above.
Converts a multidimensional research field into a marketing slogan. Greys have the richest symbolic-cognition literature; crows are outstanding tool makers; Goffin's are extraordinary innovators; kea excel at exploratory physical cognition.
The viral side of talking birds — Disco, the several Einsteins, and edited-clip evidence generally — is on famous talkers. How a budgie can build a repertoire like Puck's is in budgie talking ability and vocal learning. The experimental caveats behind every number here are on cognition.