At which OSI layers does Ethernet operate, and what IEEE standards define it?

Question

Accepted Answer

The Data Link Layer (L2) and the Physical Layer (L1); defined by IEEE 802.2 (LLC sublayer) and IEEE 802.3 (MAC sublayer plus physical signalling).
Ethernet is unusual in that it spans the bottom two OSI layers rather than just one. At the Physical Layer (Layer 1) it specifies the cabling, connectors, and how bits become electrical or light signals; at the Data Link Layer (Layer 2) it defines how those bits are framed and addressed. That is why an "Ethernet standard" name like 1000BASE-T describes both the wire and the framing.
Two IEEE working groups split the job:

IEEE 802.2 defines the LLC (Logical Link Control) sublayer — the upper part of Layer 2 that identifies which network-layer protocol the frame carries (e.g. IPv4 vs IPv6).
IEEE 802.3 defines the MAC (Media Access Control) sublayer plus the Physical Layer signalling — addressing, framing, and putting bits on the wire.

This layering is why Ethernet has been the dominant wired LAN technology since the 1990s: the same Layer 2 framing rides over many different Layer 1 media (copper, fiber) without the upper layers needing to care.
Go deeper:

Ethernet (Wikipedia) — overview of the standard family, its history, and how it maps to the OSI physical and data-link layers.

IEEE 802.3 (Wikipedia) — the working group defining the MAC sublayer and physical layer, from 10BASE5 to 400-Gigabit Ethernet.