Stable Diffusion 3.5 Medium

Pricing — 1 offering(s)

Showing the active price and any recorded history. Full pricing history is available via the paid API — see API docs.

Capability profile

Operator guidance

Route here when budget and throughput matter more than peak fidelity; step up to SD 3.5 Large (or FLUX.1 [dev]/[pro]) for higher quality.

Use cases

• Cost-sensitive, higher-volume generation where Large's fidelity isn't needed
• Local / self-hosted use (open weights, smaller footprint than Large)

Limitations

• Lower fidelity and detail than SD 3.5 Large
• No separately published GenEval/T2I-CompBench scores tracked yet. An independent Elo does now exist: Artificial Analysis Text to Image Arena (artificialanalysis.ai/image/leaderboard/text-to-image, confirmed 2026-08-19, SCO-462 sweep) ranks 'Stable Diffusion 3.5 Medium' #132 of ~147 models tracked, Elo 963 — below both Large variants, consistent with its smaller ~2.5B size. A different benchmark type than GenEval/T2I-CompBench, not a substitute for one

Citations

• Stable Diffusion 3.5 (Stability AI announcement)
• Stable Diffusion 3.5 Medium — Artificial Analysis Text to Image Arena (independent Elo)
• Stable Diffusion 3.5 Medium on Replicate