Kimi K3.1 Pro Launches: Model Upgrade, User Quota Dispute Emerges: what are the key facts?

In the early hours of October 24, Moonshot AI released the Kimi K3.1 Pro model, with parameters reaching the 3-trillion scale and support for a 1-million-token context.

Kimi K3.1 Pro Launches: Model Upgrade, User Quota Dispute Emerges: what are the key facts?

The model scored 65.7% on the Terminal-Bench 4.0 test, approaching the level of Claude Opus 5.5.

Kimi K3.1 Pro Launches: Model Upgrade, User Quota Dispute Emerges: what are the key facts?

K3.1 Pro's API pricing is 13 yuan per million tokens for input and 68 yuan for output, lower than the previous generation K3's 20 yuan and 100 yuan.

Kimi K3.1 Pro Launches: Model Upgrade, User Quota Dispute Emerges: what are the key facts?

Some users report that the 199-yuan Pro plan's quota is insufficient, and one developer's test claims the old plan's API value is about 1.94 times that of the new plan.

Kimi K3.1 Pro Launches: Model Upgrade, User Quota Dispute Emerges: what are the key facts?

A China Academy of Information and Communications Technology report shows that in June 2026, domestic average daily token consumption was about 500 trillion, nearly 4 times that of December 2025.

Physix Frontier · News Briefing Card (enbrief · Oct 25, 2026)

Kimi K3.1 Pro Launches: Model Upgrade, User Quota Dispute Emerges

KEY FACTS

In the early hours of October 24, Moonshot AI released the Kimi K3.1 Pro model, with parameters reaching the 3-trillion scale and support for a 1-million-token context.
The model scored 65.7% on the Terminal-Bench 4.0 test, approaching the level of Claude Opus 5.5.
K3.1 Pro's API pricing is 13 yuan per million tokens for input and 68 yuan for output, lower than the previous generation K3's 20 yuan and 100 yuan.
Some users report that the 199-yuan Pro plan's quota is insufficient, and one developer's test claims the old plan's API value is about 1.94 times that of the new plan.
A China Academy of Information and Communications Technology report shows that in June 2026, domestic average daily token consumption was about 500 trillion, nearly 4 times that of December 2025.

KEY DATA

3-trillion scaleModel Parameters

1 million tokensContext Length

65.7%Terminal-Bench 4.0 Score

13 yuan/68 yuan per million tokensAPI Price (Input/Output)

PHYSIX OBSERVATION

Model capability improvements and API price cuts are good things, but the perceived rise in users' actual usage costs shows a mismatch between subscription models and Agent task consumption. The industry needs to shift from selling tokens to selling work outcomes; otherwise, compute pressure will continue to be passed on to users.

Source: enbrief original report ↗