> ## Content Index > Fetch the complete content index at: https://www.testingcatalog.com/llms.txt > Use this file to discover other available public pages before exploring further. # Kimi K3.1 is prepared for upcoming release after Arena testing - URL: https://www.testingcatalog.com/kimi-k3-1-spotted-testing-arena/ - Published: 2026-10-23T09:00:38.000Z - Updated: 2026-10-23T09:01:03.000Z - Description: A mystery model called Kibitz has turned up on Arena, and testers believe it's Moonshot AI's Kimi K3.1. Early outputs look strong, especially for 3D work, and a launch may be days away. - Author: Dwaybe Cubbins - Tags: Kimi, AI Rumours, Latest AI News Moonshot may be close to dropping its next major model. Many users have noticed a new anonymous model called **Kibitz** on Arena, and most speculate it could be Moonshot's Kimi K3.1\. Kimi also published a teaser for the upcoming Kimi K3.1 launch, and the Arena account on X sent a similar message. > [pic.twitter.com/Ev0g2EbAyw](https://t.co/Ev0g2EbAyw?ref=testingcatalog.com) > > — Kimi.ai (@Kimi\_Moonshot) [October 23, 2026](https://x.com/Kimi%5FMoonshot/status/2113445757185606062?ref%5Fsrc=twsrc%5Etfw&ref=testingcatalog.com) > A new model from [@Kimi\_Moonshot](https://x.com/Kimi%5FMoonshot?ref%5Fsrc=twsrc%5Etfw&ref=testingcatalog.com) is coming. Stay tuned for where it lands in the Arena. [https://t.co/ZKbFEEQJsU](https://t.co/ZKbFEEQJsU?ref=testingcatalog.com) > > — Arena.ai (@arena) [October 23, 2026](https://x.com/arena/status/2113492200168579111?ref%5Fsrc=twsrc%5Etfw&ref=testingcatalog.com) Chetaslua was one of the first to spotlight Kibitz earlier today, calling it *"a new mysterious model on code arena"* and guessing it was likely from Kimi. In a follow-up post on X, they said they were impressed by Kibitz's JS drawing skills and that it was fast. They also added that they had heard from *"friends inside Kimi"* that it will be a very efficient model, though that remains secondhand. > 🚨 Kimi Kibitz JS Drawing ABilities is Top Notch > > Its one shot obviously as on arena you can not ask same ai model twice in battle mode > > This is so fast , and i heard this will be going to very efficient model from my friends inside Kimi [https://t.co/wvSO1AQ5mg](https://t.co/wvSO1AQ5mg?ref=testingcatalog.com) [pic.twitter.com/NSb1wzir2G](https://t.co/NSb1wzir2G?ref=testingcatalog.com) > > — Chetaslua (@chetaslua) [October 22, 2026](https://x.com/chetaslua/status/2113184816044949916?ref%5Fsrc=twsrc%5Etfw&ref=testingcatalog.com) > 🚨 Kimi K3.1 Pro took 15 min to one shot this demo website > > welcome to the new era of webdev [https://t.co/x5M3VS5Khg](https://t.co/x5M3VS5Khg?ref=testingcatalog.com) [pic.twitter.com/drpLtBW9NP](https://t.co/drpLtBW9NP?ref=testingcatalog.com) > > — Chetaslua (@chetaslua) [October 23, 2026](https://x.com/chetaslua/status/2113552063636578445?ref%5Fsrc=twsrc%5Etfw&ref=testingcatalog.com) Apart from that, X user Chris, who had told followers to keep an eye on Arena for Kimi this week, dubbed it Kimi K3.1 and claimed that **the model will release before Tuesday next week as a low-cost, SOTA open-source model**. The comparison Chris shared asks each model to *"build what you think you look like in Minecraft."* According to the post's figures, GPT-6 Astra placed 5,738 blocks at a cost of $5.31, while Kimi K3.1 placed 13,707 blocks for $0.38\. One thing we noticed in the video is that Kimi K3.1 built itself to look like the blue Kimi Agent mascot, while GPT-6 settled for a generic robot. It's a small detail, but it suggests the model has a clearer sense of its own brand than you might expect. > 🚨Kimi k3.1 (Kibitz on Arena) will release before Tuesday next week. > > It will be a state of the art Open Source model at low cost. > > “build what you think you look like in Minecraft” > > GPT6 ASTRA > 5,738 BLOCKS • $5.31 > > KIMI K3.1 > 13,707 BLOCKS • $0.38 [https://t.co/sfuoLWuJMo](https://t.co/sfuoLWuJMo?ref=testingcatalog.com) [pic.twitter.com/z3p1aCv4VX](https://t.co/z3p1aCv4VX?ref=testingcatalog.com) > > — Chris (@ChrisGPT) [October 22, 2026](https://x.com/ChrisGPT/status/2113180444858667485?ref%5Fsrc=twsrc%5Etfw&ref=testingcatalog.com) Other early outputs also show promising results. Lumina, for example, shared a 3D helicopter with takeoff and landing, assembly and exploded views, plus X-ray and wireframe modes, and said it took under 15 minutes to build. > Kimi K3.1 looks incredibly strong > > This is a 3D helicopter with takeoff land, 3D assembly and exploded views, X-ray and wireframe modes and a ton of detail > > This was made in under 15 minutes too [pic.twitter.com/QF5n9VjYkj](https://t.co/QF5n9VjYkj?ref=testingcatalog.com) > > — Lumina (@LuminaBench) [October 22, 2026](https://x.com/LuminaBench/status/2113186973363929470?ref%5Fsrc=twsrc%5Etfw&ref=testingcatalog.com) Another user put Kibitz up against Claude Fable 5.1 in a 3D test and found it noticeably faster than Fable, rating the result close to Opus 5.5 Medium. They suspect it runs at its highest reasoning setting by default, which would match how Moonshot shipped K3 with max effort as the default. > Kibitz (Kimi K3.1) vs Fable 5.1 on Arena, and I'm really impressed with what I'm seeing in this test! I'm guessing the new Kimi model might be using its highest reasoning setting by default because that's the only explanation I can think of for how good the 3D quality looks… [pic.twitter.com/rG8pXvymwT](https://t.co/rG8pXvymwT?ref=testingcatalog.com) > > — Iam\_ (@SPAC89) [October 22, 2026](https://x.com/SPAC89/status/2113136432936083808?ref%5Fsrc=twsrc%5Etfw&ref=testingcatalog.com) Of course, these are one-shot Arena battles, so take them with a grain of salt until there's a model card. But we cannot deny the pattern is pretty familiar since we got an [early look at Kimi K3 generations on Arena](https://www.testingcatalog.com/early-look-at-kimi-k3-generations-from-moonshot-ai-on-arena/) back in July under the codename Kivine, and the model launched within days. K3 arrived on July 16 as a 2.8-trillion-parameter model with a 1M-token context window, priced at $3 input and $15 output per million tokens, and its open weights followed on July 27. > Kimi K3.1 Medieval Voxel Castle > > This is in arena under "Kibitz" and feels identical to GPT-6.1 Sol Max with a sprinkle of Sonnet's design taste > > We all know what they have done and are in for a treat, they also fixed the speed problems [https://t.co/WA6E59sbKO](https://t.co/WA6E59sbKO?ref=testingcatalog.com) [pic.twitter.com/x4jowVjlI5](https://t.co/x4jowVjlI5?ref=testingcatalog.com) > > — Lumina (@LuminaBench) [October 22, 2026](https://x.com/LuminaBench/status/2113248472291852327?ref%5Fsrc=twsrc%5Etfw&ref=testingcatalog.com) While Kimi K3.1 might not go toe-to-toe with frontier models from Anthropic and OpenAI in every area, the pricing could make it a no-brainer for many. If Chris' cost figures hold up, Moonshot would undercut closed labs by a wide margin, just as OpenAI readies [GPT-6.1 Terra](https://www.testingcatalog.com/openai-prepares-gpt-6-1-terra-microsoft-code/) as a cheaper option and Anthropic gets [Fable 5.5 into partners' hands](https://www.testingcatalog.com/anthropic-several-models-coming-fable-5-5-reaches-partners/). A strong open model at that price would put real pressure on both.