Novita AI: three Time Limited Free LLMs graduate to paid pricing
Macaron V1 Venti, Macaron V1 Tall, and Ling 3.0 Flash moved off Novita AI's introductory $0 rate to paid per-token pricing, while a new model, Ling 3.0 Tiny, took over the free-launch slot.
Macaron V1 Venti, Macaron V1 Tall, and Ling 3.0 Flash listed at $0/M input and output under a Time Limited Free tag on Novita's serverless model catalog.
Macaron V1 Venti $1.5/M in ($0.3/M cache read) / $4.5/M out; Macaron V1 Tall $0.45/M in ($0.08/M cache read) / $2.6/M out; Ling 3.0 Flash $0.06/M in ($0.012/M cache read) / $0.18/M out. Ling 3.0 Tiny added as the new $0 Time Limited Free listing.
Novita AI’s serverless model catalog (novita.ai/en/pricing and novita.ai/en/models) confirms that three LLMs previously tagged Time Limited Free — Macaron V1 Venti, Macaron V1 Tall, and Ling 3.0 Flash — converted to paid per-token pricing by the 2026-08-11 capture, exactly the expiry the promotional tag implied but never dated. Macaron V1 Venti now bills $1.5/M input ($0.3/M cache read) and $4.5/M output; Macaron V1 Tall bills $0.45/M input ($0.08/M cache read) and $2.6/M output; Ling 3.0 Flash bills $0.06/M input ($0.012/M cache read) and $0.18/M output.
A new small model, Ling 3.0 Tiny, was added at $0/M flat under the same Time Limited Free tag, continuing Novita’s pattern of using a free listing as a launch ramp for a new model rather than a permanent tier — the same mechanic seen when Tencent’s Hy3 graduated off free pricing on 2026-07-21. The overall catalog held steady at 177 models, and this capture cycle found no other rate changes: GPU instances, dedicated endpoints, bare-metal nodes, and Agent Sandbox pricing were all byte-identical to the 2026-08-04 capture.
Macaron V1 Venti, Macaron V1 Tall, and Ling 3.0 Flash moved off their introductory Time Limited Free tag to paid per-token rates (Macaron V1 Venti $1.5/M in · $4.5/M out; Macaron V1 Tall $0.45/M in · $2.6/M out; Ling 3.0 Flash $0.06/M in · $0.18/M out), while a new small model, Ling 3.0 Tiny, was added as the new $0 free-launch listing under the same tag. Catalog held at 177 models; GPU, dedicated-endpoint, bare-metal, and Agent Sandbox rates were unchanged.