Skip to content

Coming to CheapestInference — models under review

New models don’t land in a pool by surprise. Every candidate goes through the same review — Reviewing → Confirmed → Capacity secured → Live — and this page is its live status board. Check back or watch the changelog for the moment a model goes live.

ModelStatePool candidateNotesUpdated
Kimi K3.1 ProConfirmedFlagshipMoonshot’s post-trained K3, released 2026-10-23 — 55 on the AA index (v4.3.2), #4 overall and 3 points off the leader, at a $2/$10 list price; 1M context, image + video input. Open weights announced for 2026-11-04; goes live once the weights and license are published and capacity is validated. Kimi K3 stays served meanwhile. Analysis2026-10-26
MiMo-V2.6-ProReviewingTBDXiaomi’s 1.02T/42B-active MoE — the top open-weights score on the AA index (46, v4.3.2) at $0.87/M output, MIT weights, omnimodal, 1M context. Quality and fit under evaluation. Analysis2026-09-22
GLM-5.3-FlashReviewingTBDZ.ai’s 320B/18B-active MoE — 42 on the AA index (v4.3.2) at $0.10/M blended, MIT weights, native multimodal, 1M context. Quality and fit under evaluation. Analysis2026-09-22

Models that completed this pipeline and went live:

  • 2026-09-23 — MiMo V2.6 Flash — upgraded the Core Pool’s MiMo slot in place, two days after Xiaomi published the open weights
  • 2026-09-10 — DeepSeek V4.1 Flash — upgraded the Core Pool’s DeepSeek slot in place, the day DeepSeek published the open weights
  • 2026-08-30 — GLM 5.3 — upgraded the Frontier Pool’s GLM slot in place, two days after Z.ai published the open weights
  • 2026-08-14 — Qwen3.8 Max — joined the Flagship Pool next to Kimi K3, the day after its open-weight variant shipped and lifted the licensing gate
  • 2026-07-31 — DeepSeek V4 Flash (0731 build) — upgraded in the Core Pool
  • 2026-07-28 — Kimi K3 — launched the Flagship Pool
  • Full history in the changelog

Tell us what you’d pay a flat monthly fee for: support@cheapestinference.com. Real demand moves models up this list.