How Many Gods? — the 33-koti count and the "330 million" mistranslation

The Bṛhadāraṇyaka Upaniṣad stages the question as a direct exchange. Vidagdha, son of Śakala, asks the sage Yājñavalkya, plainly: kati devāḥ — "how many gods are there?" It's a debate scene, not a devotional aside, and Yājñavalkya's first move is a joke at the questioner's expense: he starts by citing the number given in a liturgical formula — 3,306 — before narrowing it down, in successive questions, to 303, then 33, then 6, then 3, then 2, then one and a half, and finally to one. It's worth reading that sequence for what it actually is: a philosophical demonstration that any large ritual enumeration collapses to a smaller functional one, and finally to Brahman itself, not a scripture asserting that Hinduism has thousands of gods and then hedging.

The number that matters for the popular claim this essay is fact-checking is the one in the middle of that sequence — 33 — because that's the number Yājñavalkya actually explains in detail rather than passing through.

Yājñavalkya's own breakdown (Bṛhadāraṇyaka Upaniṣad 3.9.1–2): the eight Vasus, the eleven Rudras, and the twelve Ādityas — thirty-one — plus Indra and Prajāpati, making thirty-three. He goes on to gloss the Vasus themselves as Fire, Earth, Air, the Intermediate Space, Sun, Heaven, Moon, and the Stars — cosmic phenomena treated as the same category as "gods," which is itself a useful corrective to reading deva as a one-to-one match for the English word "god."

This isn't a late invention read backward into the Vedas. The Ṛgveda Saṃhitā itself invokes the number directly — RV 1.45.2 calls on "the three and thirty divinities," a clean citation of the number. RV 1.34.11 is a softer one, worth citing precisely rather than folding into the same basket: it invites the Aśvins to a rite with "thrice-eleven gods" (33, by multiplication, but framed as the Aśvins' own retinue in that specific hymn, not a general census of the pantheon).

Here's the honest complication, and it's worth stating rather than smoothing over: the number 33 is stable across sources, but the membership list is not.

• Bṛhadāraṇyaka Upaniṣad's own list: 8 Vasus + 11 Rudras + 12 Ādityas + Indra + Prajāpati.
• The Brāhmaṇas' standard enumeration: 12 Ādityas + 11 Rudras + 8 Vasus + the two Aśvins — no Indra or Prajāpati in this version at all.
• A Śatapatha Brāhmaṇa variant: Heaven and Earth folded in among the Vasus, with Indra counted as the 32nd god and Prajāpati as the 33rd.

None of these is "the wrong one" that later scribes corrupted — they're parallel enumerations from different ritual contexts, and the fact that Vedic literature can hold a stable number with an unstable membership list is itself a real, citable feature of how this material was transmitted, not a contradiction to paper over.

One popular framing worth addressing directly: avatāras are not part of any version of this list. Rāma, Kṛṣṇa, and the rest of Viṣṇu's avatāras belong to a separate category (avatāra, "descent") layered on top of the Vedic deva framework centuries later in the Purāṇic material — none of the Vedic or Brāhmaṇa enumerations of the 33 includes them. Where the Aśvins do show up (the Brāhmaṇa version above), that's a real textual variant worth knowing, not a mistake.

The "330 million" figure comes from a real word, misread. Koti carries two genuinely distinct senses in Sanskrit: in ordinary modern usage across Hindi and other Indian languages, koti/crore means the number ten million. But in older and more technical liturgical usage, koti also means "type," "category," or "supreme/excellent" — a sense well enough attested that when Tibetan translators rendered Sanskrit Buddhist and Hindu texts into Tibetan, they translated koti in this context with rnam, meaning "class" or "kind," not with a Tibetan numeral.

"Trayastriṃśat-koti devāḥ" — "the thirty-three koti of gods" — was therefore originally a statement about 33 types or classes of divine being (the Vasus, Rudras, Ādityas, and so on, as categories), not a headcount. Once the numeral sense of koti became the dominant one in vernacular usage — a gradual drift over centuries, not a single colonial mistranslation event, though colonial-era and popular Purāṇic-period retellings certainly repeated and cemented it — "33 koti" was read as "33 crore," and 33 crore is 330 million. The number that gets repeated today as "Hinduism has 330 million gods" is a numeral reading of a word that, in its original context, wasn't giving a headcount at all.

A specific claim worth checking rather than assuming: some framings circulating today cite a figure of 32,000 gods. Having gone looking for it specifically — across the primary texts above, the classical commentarial tradition, and even the popular (and often inaccurate) online explainers that repeat the "330 million" figure — no scriptural or traditional source for "32,000" turned up anywhere. The numbers that genuinely recur across sources cluster around 33 (the Vedic/Upaniṣadic figure), 33 koti in its original categorical sense, or 330 million as the popular numeral misreading. "32,000" doesn't fit any of those patterns and doesn't trace to an identifiable text. The honest answer is not to quietly substitute a number that sounds plausible — it's that this specific figure appears to be a modern, ungrounded claim, and it should be named as such rather than left unaddressed.

Strip away both the popular multiplication and the popular skepticism, and what the tradition's own core texts commit to is narrow and specific: 33, named directly in the Ṛgveda Saṃhitā and explained in full in the Bṛhadāraṇyaka Upaniṣad, with real, acknowledged variation in exactly which 33 — Aśvins in one enumeration, Indra and Prajāpati in another. "330 million" is a real phrase in the tradition's own liturgical Sanskrit (trayastriṃśat-koti), but it was never a census; it's 33 kinds of divinity, not 33 crore individual deities. And the far larger, more sprawling pantheon most people are actually gesturing at when they say "Hinduism has millions of gods" — every regional deity, village guardian, river goddess, and household devatā venerated across the subcontinent's living traditions — is a real, genuinely vast and still-growing category, but it was never counted, and the tradition's own texts never claimed to count it. The precise, countable number the scriptures actually give is 33.

• Primary scripture: Bṛhadāraṇyaka Upaniṣad 3.9.1–2 (Yājñavalkya's full enumeration, including the joke-sequence from 3,306 down to one); Ṛgveda 1.45.2 ("three and thirty divinities"); Ṛgveda 1.34.11 ("thrice-eleven gods," in its narrower Aśvin-invocation context).
• Classical commentary/parallel enumeration: the Brāhmaṇas' 12 Ādityas + 11 Rudras + 8 Vasus + 2 Aśvins list; the Śatapatha Brāhmaṇa's Heaven-and-Earth-as-Vasus variant — cited to show real internal variation, not resolved into one "correct" list.
• Etymology: koti's dual sense (numeral "crore" vs. categorical "type/supreme"), cross-checked against the Tibetan translation tradition's rendering of the term as rnam ("class/kind") in this context.
• Fact-check, explicitly negative result: no traceable scriptural, commentarial, or even popular-explainer source for a "32,000 gods" figure — reported as a genuine absence rather than silently omitted.
• Excluded: modern popular explainers that assert "330 million" as a literal historical census figure without engaging the koti ambiguity at all — these repeat the mistranslation rather than sourcing it.