MA
Did Elon catch up? (Grok 4.7 is here)Matthew Berman22 Sept 2026
Open-weights model ranked just below GLM 5.3 Max on the index; shown in a third-party demo beating Grok 4.7.
KK
Kimi K3
by Moonshot AI
Moonshot AI's Kimi K3 model family, with K3.1 Pro as the current release in the Kimi app and API.
Updated EU status checked Last shown
Kimi K3 at a glance
- What is Kimi K3?
- Kimi K3 is a model from Moonshot AI. Moonshot AI's Kimi K3 model family, with K3.1 Pro as the current release in the Kimi app and API.
- Who makes Kimi K3?
- Kimi K3 is made by Moonshot AI.
- Does Kimi K3 work in the EU?
- Unverified. No explicit evidence found. Last verified 14 Nov 2026.
- How much does Kimi K3 cost?
- Pricing unknown. No pricing found in the provided material; see platform.moonshot.ai for API pricing.
- Where was Kimi K3 shown?
- In 15 creator videos by Matthew Berman, AI Search, Matt Wolfe. Most recently by Matthew Berman on 22 Sept 2026, at 13:55 in “Did Elon catch up? (Grok 4.7 is here)”.
Sources: the vendor's own pages (linked above) and creator videos at the linked timestamps. EU status is information, not legal advice; see how we check.
About
Kimi K3 is a large language model from Moonshot AI, accessed through the Kimi chatbot (kimi.com) and the Moonshot API platform. The official site currently promotes K3.1 Pro, an update that targets coding, agentic tasks, 3D model building and financial research.
A YouTube review (Matthew Berman) describes K3 as an open-weights model ranked just below GLM 5.3 Max on an index, and shows a third-party demo in which it outperformed Grok 4.7. Those claims come from a single creator and have not been verified against the vendor's pages. No weights repository, license text or technical paper was found in the provided material.
Exact parameter counts, context length, pricing and the first release date of K3 are not stated in the provided sources, so they are left unspecified here.
Key features
- Coding and complex software project support
- Agentic task execution
- 3D model generation, according to the vendor
- In-depth financial research
- Chat app and developer API access
15 moments
As seen on YouTube
Watch the moment: the exact second top creators show it. Videos load only when you press play, via youtube-nocookie.com.
AI
Claude Fable 5.1 is savageAI Search3 Sept 2026
Open model compared on cost and the brain tumor test, where it got one of six correct like GLM 5.3.
MA
The Most Important Chart In AI Right NowMatthew Berman26 Aug 2026
Open weights model that Harvey fine-tuned into a legal expert with top legal benchmark results. Half the per-token price of GPT-5.6 Sol, but completed-task cost is similar (84 cents vs 96 cents) because it uses more tokens.
AI
New Deepseek, Seedance 2.5, Minimax H3, Gemini Robotics, AMD models: AI NEWSAI Search2 Aug 2026
Moonshot released the open weights for its 2.8T-parameter MoE (104B active) with native vision, Kimi Delta Attention and attention residuals. Full weights are 1.56 TB, and the Unsloth 1-bit GGUF is 594 GB.
MA
GPT-5.6 just made itself better...Matthew Berman31 Jul 2026
Newly released Chinese open-source model near the frontier. It is half the price of GPT-5.6 Soul per token but uses about twice as many tokens per task, so the effective cost is similar.
MA
I'm disappointedMatthew Berman29 Jul 2026
Open-source Chinese model from Moonshot AI that on many measures rivals closed models. On the Artificial Analysis leaderboard it sits right behind the top Claude and ChatGPT models.
AI
Claude Opus 5 is a freakAI Search27 Jul 2026
Frontier comparison model, also capable of the financial video task, cheaper than Opus 5 with similar benchmark scores.
MA
What did Anthropic do?! (Opus 5)Matthew Berman24 Jul 2026
Open-source model that costs half as much per token as GPT-5.6 Soul but uses twice the tokens, so cost per task is similar. It is one of three models on the Pareto frontier alongside Opus 5 and GPT-5.6 Soul.
AI News: Kimi K3 Has Big AI Labs Panicking!
Open-weight frontier model that matches or beats GPT 5.6 Soul and Fable 5 on several benchmarks. Amid distillation accusations, the creator tested it via the Kimi playground and Kimi Code, where it built a Megabonk clone in one prompt, and on Busy Bench it ranked seventh.
MA
You NEED to do this (HUGE AI SAVINGS)Matthew Berman23 Jul 2026
Chinese open-source model priced at $3 per million input and $15 per million output tokens. It is competitive on benchmarks, but it uses about twice as many tokens, so the cost per task is similar to GPT-5.6.
MA
The Most Important Conversation in AI Right NowMatthew Berman21 Jul 2026
Open-weight frontier model with 2.8T parameters, 1M token context and native multimodality, comparable to top OpenAI and Anthropic models. Priced about $3/M input and $15/M output tokens, but uses roughly double the tokens, so cost per task matters. Discussed amid US talk of restricting Chinese open models.
AI
Kimi K3, dancing waifus, robot UFC, song to MIDI, GPT Red, hoverboards: AI NEWSAI Search19 Jul 2026
Top open-weights model matching frontier models on many benchmarks. New leaderboard results: 2nd on debate, beats Opus 4.8 on KernelBench CUDA, 69% on DeepSWE, but lower on LiveBench; also designed a chip autonomously over 48 hours.
This model just changed everything...
Moonshot AI's 2.8 trillion parameter open-weights model with a 1M token context window, aimed at long-horizon coding, knowledge work and reasoning. It ranks first on Arena AI's front-end benchmark (76% vs 63%) and on Next.js evals, and is priced at $3/$15 per million tokens. It is token-hungry and slow, and the creator's Rubik's Cube simulator build took about 30 minutes but worked.
MA
AI News: Claude's New Browser, Spotify Gets AI & OpenAI's New HardwareMatt Wolfe17 Jul 2026
2.8 trillion parameter model with weights promised this month; benchmarks beat GPT 5.5, Opus 4.8 and GLM 5.2 and trail only the top 5.6 and Fable models. The creator hasn't tested it yet.
New #1 open source AI model is here! Kimi K3
Moonshot AI's new 2.8T-parameter model with native vision and 1M context, with weights to be released by July 27th. The creator tests it on physics sims, Blender, video generation, a piano, a game, music, medical images and deep research. Benchmarks put it close to GPT 5.6 and Claude Fable 5 at lower cost. It is slow, and it failed the frog test and most of the cancer scan test.
Keep exploring
Similar AI tools
Closest neighbours by meaning, not just by category.
KK
Kimi K3.1
Model · Moonshot AI
Moonshot AI's open-weight 2.8T-parameter multimodal MoE agentic model with 1M-token context.
Available in EULanguage Models
2 Nov 2026API
K3
Kimi 3 Max
Model · Moonshot AI
Moonshot AI's large language model, described as open-source with roughly 2T+ parameters, benchmarked by Artificial Analysis.
UnverifiedLanguage Models
Release date n/a
KK
Kimi K2.5
Model · Moonshot AI
Moonshot AI's open-weights 1T-parameter MoE multimodal agentic model with 256K context, vision input and agent swarm.
Available in EULanguage Models
Jan 2026Open source
KK
Kimi K2.7 Code
Model · Moonshot AI
Moonshot AI's Kimi K2.7 coding model, offered through the Kimi Code agent (CLI, VS Code extension, desktop app).
UnverifiedLanguage Models
Release date n/aPaid
KW
Kimi Work
App · Moonshot AI
Moonshot AI's desktop app that lets Kimi models work with files on a user's local computer.
UnverifiedProductivity
Release date n/a
KC
Kimi Code
Agent · Moonshot AI
Moonshot AI's terminal coding agent and VS Code extension, powered by Kimi models; successor to the archived Kimi CLI.
UnverifiedCoding
15 Oct 2025Open source