The story of Şúnaşşépa from the Āitareya-brāhmaná indicates an unambiguously literal man-sacrifice, and is attested in the Sáṁhitā itself via the hymns of the R̥g-vedá authored by Şúnaşşépa, which make references to his impending doom, as well as R̥g-vedá 5.2.7. The literal man-sacrifice also survived well into the Sū́tra æra, with both the Şāṅkhāyana-şrāuta-sūtrá (16.13.7) and the Vāitāna-şrāuta-sūtrá (7.2.26) explicitly mentioning a suffocated victim in their coverage of the puruṣa-medhá.
A man was originally sacrificed not only at the puruṣa-medhá but uncontroversially at several other rites like the Agni-cáyana in Kātyāyana-şrāuta-sūtrá 16.1.14 +++(परिवृते पुरुषसञ्ज्ञपनम् १४)+++, Şúnaskárṇa’s Agni-ṣṭomá in Pañca-vin̐şa-brāhmaṇá 17.12.5, and probably at the aşva-medhá as mentioned in Tāittirīya-brāhmaṇá 3.9.8.
The Sū́tra-s that don’t mention a literal man-sacrifice generally don’t specify a symbolic one, either—the default implication, euphemistically passed over by those texts, would be a killing of the victim. The only exception of which I’m aware is the Vādhūla-şrāuta-sūtrá, which specifies the freeing of the victim.
Famously the Şata-patha-brāhmaṇá supposedly “condemns” a literal man-sacrifice (13.6.2.13). Yet the very same text also mentions explicitly the slaughter of a man in the Agni-cáyana (6.2.1.18). If the Şata-patha-brāhmaṇá did condemn man-sacrifice, it would have to be taken as an aberration, not a proof that man-sacrifice was metaphorical in Vedic times; but in reality, a close reading of the passage shows that there is no such condemnation. The passage mentions an instance (in the past tense) of the freeing of victims because sacrificing them could result in cannibalism (since priests partake in the offering), and isn’t an injunction against killing the victims in general—only against cannibalism.