PyyanThe AI Index

Pyyan / Compare / Kimi K3.1 Pro vs Gemini 4 Argon

Kimi K3.1 Pro vs Gemini 4 Argon

2 of 5

Language Models · verified 26 Oct 2026

Kimi K3.1 ProMoonshot AIcurrent
Gemini 4 ArgonGooglecurrent
3 slots left
SpecificationKimi K3.1 ProGemini 4 Argon
SummaryFrontier coding numbers at a fifth of the price, with the weights promised for November.First on one independent board, fifth on another, and cheap until December.
Context1M1M
Input ($/Mtok)$2$2
Output ($/Mtok)$10$10
LicenceClosed for now, weights promised 4 Nov 2026Proprietary
Max output944K262K
Parameters3T class, exact count not publishedNot published
Released23 Oct 202630 Sep 2026, GA 13 Oct
Model IDkimi-k3.1-progemini-4-argon
CategoryLanguage ModelsLanguage Models
OfficialMoonshot AI ↗Google ↗

Highlighted rows are where these differ.

Kimi K3.1 Pro

  • 65.7% on Terminal-Bench 4.0 in Moonshot's own table, against 66.4% for Claude Opus 5.5 and 58.2% for GPT-6 Astra
  • Claim-level hallucination rate down from 23.4% on K3 to 8.3%
  • Cyber training data deliberately removed, which Moonshot says costs about nine points on xdev-bench-hard
  • Full weights promised by 4 November 2026, so the licence is not yet knowable

Best for long-horizon coding when the budget is the constraint.

Full spec sheet →

Gemini 4 Argon

  • The price doubles to $4 and $20 on 1 December, and cached input is discounted 95%
  • First of 41 on the Vals Index at 68.90%, but 52.6 on Artificial Analysis against 57.6 for Opus 5.5
  • 77.9 on DeepSWE v1.1 in Google's table, and 55.0 on FrontierSWE v2 where GPT-6 Astra scores 65.5

Best for frontier work while the introductory rate lasts.

Full spec sheet →