Dana retweeted
Most interesting observation from new AA V5 benchmark:
Despite having private benchmarks that can not be benchmaxed Kimi 3.1 Pro (46) is better than Muse Wonder (43) and right behind Fable 5.1 (49).
This should put to rest most doubts how good this @KimiDevs , @Kimi_Moonshot model is; and why I say it was such a responsible thing for them to reduce it cyber capabilities.
@teortaxesTex
Announcing Artificial Analysis Intelligence Index v5, featuring 2 new benchmarks and 3 upgrades. For the first time, the majority of the Index is built on private datasets
The Artificial Analysis Intelligence Index is our composite measure of model capabilities across tasks reflecting real-world usage today. v5 is our biggest upgrade since v4 launched in January. With v5, 60% of the Index is now based on evaluations with private test sets or held-out answers.
Alongside Intelligence Index v5, we’re launching the Artificial Analysis Progress Index v1, which tracks model capability over time on one continuous scale.
New to the Intelligence Index:
➤ FrontierCode 2.0: Cognition's newly-launched private benchmark of real-world coding tasks, graded on whether a repository maintainer would merge the agent's change
➤ Terminal-Bench-Science: 70 agentic scientific research tasks, run in a terminal environment
Upgraded:
➤ HLE-Diamond (replacing Humanity's Last Exam): a refined 1,000-question version of Humanity's Last Exam, split evenly between reasoning and knowledge questions
➤ AA-Omniscience v1.1: An updated dataset with saturated and public set questions removed, new questions added, and ambiguous or erroneous questions revised
➤ CritPt v2: a private, 40-question hard subset of CritPt v2, revised by the CritPt authors and run with a code-execution tool
Staying in the Index: AA-Briefcase, AutomationBench-AA, GDP.pdf and Terminal-Bench 4.0
Removed from the Index: GDPval-AA v2.1, SciCode, and AA-LCR v1.1
Key results:
➤ Claude Opus 5.5 (max) leads at 58, ahead of GPT-6 Astra (max, 57) and GPT-6.1 Sol (max, 55): Anthropic and OpenAI hold the top five models. Gemini 4 Argon (high) is next at 50
➤ The open weights gap widens from 3 points to 12: Kimi K3.1 Pro (weights not yet released) is the top open weights model at 46. Among released weights, DeepSeek V4.1 Flash 1017 (max) and GLM-5.3 (max) lead at 32 - 26 points behind the leader and below Claude Haiku 5.5 (max, 36)
➤ Cost per task varies by more than 13x among the leading models: of the 10 highest-scoring models shown, GPT-6.1 Terra (max) is the cheapest at $0.96 per Intelligence Index task and Claude Fable 5.1 (max) the most expensive at $13.04
Dana retweeted
Kimi K3.1 Pro enters MathArena as the best open model (once they release the weights)! Its estimated overall performance is now similar to GPT-5.5 (April 24), putting the open frontier around 6 months behind on MathArena
Dana retweeted
We're sorry for the rate limits and unnecessary quota consumption on the K3.1 Pro subscription plan. These were our issues, and we should have fixed them sooner.
Today, every subscriber on every plan gets a one-time usage reset, restoring 25% of their monthly quota (usage below 25% resets to zero), equal to one week of usage. Your usage bar will reflect this adjustment by end of day PST.
We've also adjusted the plan's parameters and are reducing unnecessary subagent calls and output length, so your quota goes further.
Thank you for your patience while we put this right.
Dana retweeted
BREAKING: Kimi K3.1 Pro by @Kimi_Moonshot is projected to rank #1 among open-weights models on DesignArena's Website leaderboard, and is tied for #3 among all models.
Kimi K3.1 Pro has a 15 Elo point lead on their previous model, Kimi K3, on the open weights leaderboard. Among all models it ranks behind only Muse Wonder (max) and Muse Spark 1.3 (xhigh). That ties it with Muse Spark 1.3 Max for third place on the overall Website leaderboard.
Congrats to the @Kimi_Moonshot team on the launch!
Dana retweeted
Kimi K3.1 Pro jumps 12 places over K3 to #4, at 2/3 of K3's price.
Kimi K3.1 Pro launches at 4th on the Artificial Analysis Intelligence Index scoring 55, making large gains in agentic performance; it will be the leading open-weights model once its weights are published
@Kimi_Moonshot's new reasoning model gains 11 points over Kimi K3 (max, 44). It trails only Claude Opus 5.5 (max with fallback, 58), Claude Sonnet 5.5 (max with fallback, 56), and Muse Wonder Full (max, 56) on the Intelligence Index.
Once the weights for K3.1 Pro are released it will be the highest-scoring open weights model and the gap between the leading open weights and proprietary models on the Intelligence Index will narrow from 12 points to 3.
Key results:
➤ Large gains in agentic performance: Kimi K3.1 Pro scores 3rd on GDPval-AA with 1,846 Elo, up 312 Elo from Kimi K3, behind only Claude Opus 5.5 (1,866 Elo) and Muse Wonder Full (1,857 Elo). Kimi K3.1 Pro also achieves the highest score of any model on Terminal-Bench 4.0. K3.1 Pro scores 67%, up from 13% for Kimi K3 and ahead of Claude Sonnet 5.5 (64%). AutomationBench-AA, which tests agentic SaaS workflows, rises from 58% to 73%.
➤ Gains in other evals are more moderate: Humanity's Last Exam rises from 47% to 52%, CritPt from 23% to 25%, and SciCode from 59% to 61%. AA-Omniscience Index rises from 20 to 30, as the hallucination rate falls from 53% to 44% and accuracy rises from 48% to 51%.
➤ Cost per Task roughly equal to Kimi K3 at $1.98 per Task: K3.1 Pro uses 55K output tokens per Intelligence Index task, 14% more than Kimi K3 at 48K, while its first-party token prices are 33% lower.
Additional model details:
➤ Context window: 1M tokens.
➤ Size: 2.8T total, 104B active parameters
➤ Input modalities: Text and image
➤ Weights and license: Weights currently not published, Moonshot is expected to release the weights soon after launch
➤ Pricing: $2/$10/$0.20 per 1M input/output/cache hit tokens, 33% lower than Kimi K3 on all three rates.
Congratulations to the team at @Kimi_Moonshot on the release!
step by step, pushing the open intelligence frontier
Kimi K3.1 Pro launches at 4th on the Artificial Analysis Intelligence Index scoring 55, making large gains in agentic performance; it will be the leading open-weights model once its weights are published
@Kimi_Moonshot's new reasoning model gains 11 points over Kimi K3 (max, 44). It trails only Claude Opus 5.5 (max with fallback, 58), Claude Sonnet 5.5 (max with fallback, 56), and Muse Wonder Full (max, 56) on the Intelligence Index.
Once the weights for K3.1 Pro are released it will be the highest-scoring open weights model and the gap between the leading open weights and proprietary models on the Intelligence Index will narrow from 12 points to 3.
Key results:
➤ Large gains in agentic performance: Kimi K3.1 Pro scores 3rd on GDPval-AA with 1,846 Elo, up 312 Elo from Kimi K3, behind only Claude Opus 5.5 (1,866 Elo) and Muse Wonder Full (1,857 Elo). Kimi K3.1 Pro also achieves the highest score of any model on Terminal-Bench 4.0. K3.1 Pro scores 67%, up from 13% for Kimi K3 and ahead of Claude Sonnet 5.5 (64%). AutomationBench-AA, which tests agentic SaaS workflows, rises from 58% to 73%.
➤ Gains in other evals are more moderate: Humanity's Last Exam rises from 47% to 52%, CritPt from 23% to 25%, and SciCode from 59% to 61%. AA-Omniscience Index rises from 20 to 30, as the hallucination rate falls from 53% to 44% and accuracy rises from 48% to 51%.
➤ Cost per Task roughly equal to Kimi K3 at $1.98 per Task: K3.1 Pro uses 55K output tokens per Intelligence Index task, 14% more than Kimi K3 at 48K, while its first-party token prices are 33% lower.
Additional model details:
➤ Context window: 1M tokens.
➤ Size: 2.8T total, 104B active parameters
➤ Input modalities: Text and image
➤ Weights and license: Weights currently not published, Moonshot is expected to release the weights soon after launch
➤ Pricing: $2/$10/$0.20 per 1M input/output/cache hit tokens, 33% lower than Kimi K3 on all three rates.
Congratulations to the team at @Kimi_Moonshot on the release!
1
1,478
Dana retweeted
Keep pushing the frontier of open weight models
Introducing Kimi K3.1 Pro: our most capable 3T-class open-weight model.
🔹 1M-token context, 2.8T total / 104B active params (MoE), Multimodal.
🔹 Full-project coding: games in Unreal, Unity & Godot.
🔹 Excels at parametric 3D modeling in Blender & CAD.
🔹 More reliable at finance work: Accurate modeling with hallucination down by 65%.
🔹 Matching the strongest proprietary models across coding, agentic and knowledge benchmarks.
Kimi K3.1 Pro and K3.1 Pro Highspeed (2–3× faster) are now live on Kimi.ai, Kimi Work, Kimi Code, and the Kimi API.
Open weights by Nov 4, 2026.
🔗 API: platform.kimi.ai
🔗 Tech blog: kimi.ai/blog/kimi-k3-1-pro
k3.1 pro is here! our most capable open model 🤩
Introducing Kimi K3.1 Pro: our most capable 3T-class open-weight model.
🔹 1M-token context, 2.8T total / 104B active params (MoE), Multimodal.
🔹 Full-project coding: games in Unreal, Unity & Godot.
🔹 Excels at parametric 3D modeling in Blender & CAD.
🔹 More reliable at finance work: Accurate modeling with hallucination down by 65%.
🔹 Matching the strongest proprietary models across coding, agentic and knowledge benchmarks.
Kimi K3.1 Pro and K3.1 Pro Highspeed (2–3× faster) are now live on Kimi.ai, Kimi Work, Kimi Code, and the Kimi API.
Open weights by Nov 4, 2026.
🔗 API: platform.kimi.ai
🔗 Tech blog: kimi.ai/blog/kimi-k3-1-pro
3
834
Step 1: Solve all problems in the digital world.
Step 2: Recreate the physical world in the 3D digital world.
Step 3: ?
Introducing Kimi K3.1 Pro: our most capable 3T-class open-weight model.
🔹 1M-token context, 2.8T total / 104B active params (MoE), Multimodal.
🔹 Full-project coding: games in Unreal, Unity & Godot.
🔹 Excels at parametric 3D modeling in Blender & CAD.
🔹 More reliable at finance work: Accurate modeling with hallucination down by 65%.
🔹 Matching the strongest proprietary models across coding, agentic and knowledge benchmarks.
Kimi K3.1 Pro and K3.1 Pro Highspeed (2–3× faster) are now live on Kimi.ai, Kimi Work, Kimi Code, and the Kimi API.
Open weights by Nov 4, 2026.
🔗 API: platform.kimi.ai
🔗 Tech blog: kimi.ai/blog/kimi-k3-1-pro
two worlds co-exist. every day I get asked
"models are getting better and better. do we really need more intelligence for average human needs?"
and
"ai still can't do many things by itself. What's the next big market after coding that ai can address?"
somehow both questions are being asked with full conviction. interesting time ;)
Marc Andreessen says the gap between what AI-native startups and incumbents think AI can do is the widest he’s ever seen:
"The AI-native early stage startups have just a completely different view on what they can do with AI than the typical incumbents in almost any industry, almost without exception."
"AI is just racing ahead in capability even week by week... There are going to be a lot of industries that are really up for grabs in a fundamentally more profound way than maybe has been the case in the past."
"The founders are gonna have a level of superpowers coming from the use of these tools and these technologies that they just simply haven't had in the past... Had I had Astra and Netscape in 1994, like, oh my God. Like, Ben, what would you and I have been able to do at Netscape?"
"Those entrepreneurs are gonna color way outside the lines. We already see this. They're not gonna think of themselves as like, 'I'm just a software company,' or, 'I'm just a hardware company,' or even, 'I'm just an AI company.'"
@pmarca @bhorowitz
1
667
Dana retweeted
BREAKING NEWS🚨: NVIDIA has just achieved 423 tok/s/user 🚀 on the 2.8T Parameter Kimi K3 model via a new upstream vLLM optimization. 1.68x Faster on the Same Hardware!
This was done by NVIDIA CentML's Benjamin, who wrote NVIDIA's new decode-specialized kernels in Cute DSL via aggressive Megakernel fusion. This is a 12,000 LoC Megakernel and is the future of advanced AI-assisted kernel generation. NVIDIA is already working on DeepSeekv4.1 and other models for vLLM too.
Someone should give this NVIDIA 10x engineer a raise for this great new major feature!
iykyk
Humans are the minority user of AI
Agents burn nearly 5x the tokens people do, up 14x since February
More charts in State of Markets II: a16z.news/p/state-of-markets…
2
1,042
"skyrocket" 👀
According to data from SensorTower, Kimi app downloads nearly quintupled and daily active users jumped by ~40%. The K3 hype is translating into real users.
Charts of the Week: a16z.news/p/charts-of-the-we…
3
1,339
interesting
Should you chase hype or ignore it? The tech industry has been debating this for decades. So at @Sequoia, we dug into 20 years of hype data.
This summer, I worked with Sequoia intern @ochonaut to measure hype over the past 2 decades. The chart below ranks the most hyped topics on Hacker News for every year since 2007. Under each year sits the most valuable company founded that year.
Here are a few observations:
1/ The top company founded in a given year is rarely related to the hype of that period. Airbnb was founded in 2008, when the top topic was Google. Uber arrived in 2009, while the conversation revolved around low-level programming. Anthropic came in 2021, while the internet was consumed by crypto. Chasing hype rarely leads to enduring outcomes. The top companies of recent years have yet to be decided.
2/ New trends announce themselves five to six years early. LLMs first cracked the top 15 in 2016 and took until 2022 to hit #1. AI coding entered at #12 in 2021 and tops the list in 2026. Crypto entered in 2011 before 2017 and 2021 peaks. “New” trends don’t appear out of nowhere, and internet subcommunities are often the first to know where the puck is headed.
3/ Long-term “hype” is a durable signal. The “Musk-Verse” has been a top 15 topic for every one of the past 14 years. Sustained attention on the internet is rare and tends to mark something real.
Next up, we want to run the same analysis with sources like X and LinkedIn. If that's of interest, give us some encouragement and we'll share the results.
4
915
🙋🏻♀️
It takes a crew to land a moonshot. Join the Kimi Ambassador Program.
Anyone who's turned an untested idea into reality knows: moonshots are never a solo mission. It takes a crew.
The Kimi Global Ambassador Program finds likeminded individuals on the same path, bringing the crew together. We hope that you:
🌙 Are influential in your own community, whether it be in tech, entrepreneurship, content creation, or on campus
🌙 Have implemented Kimi K3 into products, agents, workflows, teams, and communities
🌙 Are willing to share your experiences and passions to a broader community
Apply here: kimi.com/lp/kimi-ambassador
3
812

























