Ling-3.0-flash

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 cost and serving efficiency are the overriding constraints and the task is within reach of a ~5B-active model — Ant's efficiency play. For heavier reasoning, use the sibling Ring models or a larger flagship. All capability signal is vendor-reported (Ant's own announcement / inclusionAI benchmarks); there is no first-party Ant pricing console and no independent benchmark reproduction, so treat the "matches our 1T flagship" claim as directional.

Use cases

• Production-scale agent deployments where per-token cost dominates and a small active footprint is decisive
• High-throughput coding / tool-use pipelines at near-zero token cost
• Self-hosted deployment under MIT; pair with the Ring family for reasoning-heavy steps

Limitations

• All benchmark/capability claims are from Ant / inclusionAI's own materials — no independent reproduction found
• No verifiable first-party Ant Group pricing console; distributed purely through third-party hosts (OpenRouter, Novita, Vercel AI Gateway, ZenMux)
• 1M context is an ambition, not shipped — native window is 256K
• Registry pricing moved from a $0/$0 launch promo to a third-party steady-state rate via a price-monitor proposal; re-verify against a stable source

Citations

• inclusionAI/Ling-3.0-flash (Hugging Face, MIT open weights)
• Ant Ling launch announcement (2026-07)
• Ling-3.0-flash on OpenRouter (host + pricing)