Need a China coding plan (GLM / MiniMax / Kimi / Volcengine / Xiaomi)? Email me at [email protected]
Moonshot AI · Kimi Code
Kimi K3 flagship (2.8T parameters, near-frontier coding) is live with a new K3-256K budget variant that halves consumption under 256K context. A new membership system is coming: Kimi and Kimi Code benefits will split; existing subscribers keep combined benefits, and new sign-ups are currently waitlist-only.
Kimi K3 (plus the new K3-256K) and the K2.7 Code family.
Moonshot’s flagship: 2.8 trillion parameters, up to 1M-token context with low/high/max thinking modes; officially described as the first open 3T-class model, stably ahead of Opus 4.8 and GPT 5.5. The new k3-256k gives identical results within 256K at roughly half the consumption (no video input). Unlocked from Moderato.
Previous workhorse for code completion and routine development, available to all members; the high-speed variant (~5-6x faster) unlocks from Allegretto.
Five tiers cover casual to professional. Annual billing saves up to ¥1,680. Purchase buttons are currently "reserve subscription" while the new membership system is pending.
Adagio
FreeCasual users and newcomers
Andante
¥39/mo annually (¥468/yr)
Individual developers
Moderato
Recommended · K3¥79/mo annually (¥948/yr)
Daily high-frequency coders
Allegretto
¥159/mo annually (¥1,908/yr)
Heavy agent workflows
Allegro
Top · 1M context¥559/mo annually (¥6,708/yr)
Power users and teams
Coding is only the start.
K3 / K2.7 for generation, refactoring and debugging.
Spin up site prototypes and landing pages.
Generate technical and API documentation.
Quickly build shareable presentations.
K3 is available from Moderato (¥99/mo) and above, including the new K3-256K budget variant. Adagio and Andante keep using K2.7. Allegro uniquely unlocks 1M-token conversations.
A new model ID that returns identical results to K3 within 256K context while consuming roughly half the quota (full K3 with 1M context costs about 2x). It does not support video input. Use it for daily and single-file tasks to save quota.
Purchase buttons currently read "reserve subscription". Official banner says a new membership system is coming that will split Kimi benefits from Kimi Code benefits; existing subscribers keep combined benefits, and buying before the switch locks in the current combined plan.
Quotas use a 5-hour window and refresh every 7 days; unused quota does not roll over. Maximum concurrency is 30.
Available from Moderato: Allegretto gets 40 cluster runs/mo with 4 subtasks; Allegro gets 8 subtasks and 4 parallel Agents.
Kimi CLI, Kimi Code for VS Code, Claude Code and Roo Code, all sharing the same plan quota.