Kimi has launched K3, a large multimodal open-weight model that the company says has 2.8 trillion total parameters. The model processes images and video natively and supports a context window of one million tokens. Kimi describes K3 as the first open model in the roughly 3 trillion parameter range. Full model weights are scheduled for release by July 27. The model is designed for long-running programming tasks, knowledge work, and complex reasoning.
Performance Benchmarks
In Kimi's own benchmarks, K3 trails the top proprietary models Claude Fable 5 and GPT 5.6 Sol but beats every other system tested. This includes the Claude Opus models and Chinese rival GLM-5.2. All results come from Kimi and were achieved at maximum or high thinking intensity, according to the company.
K3 wins two out of six programming benchmarks and finishes second or third in the rest. Among general agents, K3 wins three out of six tests. Fable 5 takes the top spot in both visual agent tests. Across all 35 tests, K3 took first place about seven times and landed second or third in most of the rest. Fable 5 won the most individual tests. In nearly every benchmark, K3 beat Opus 4.8, GPT 5.5, and GLM 5.2 by a wide margin. Depending on the benchmark, one of three agent systems was used: KimiCode, Claude Code, or Codex. That means the results were not all collected under identical conditions.
Independent Testing Confirms Strong Performance
Independent testing lab Artificial Analysis has published its first evaluation of Kimi K3. The model scores 57 on the Artificial Analysis Intelligence Index, putting it on par with Opus 4.8 and GPT-5.5 but still behind Fable 5 and GPT-5.6 Sol. That largely lines up with Kimi's own claims. K3 scores 57 on the Artificial Analysis Intelligence Index, placing fourth overall behind Claude Fable 5 at 60, GPT-5.6 Sol at 59, and Claude Opus 4.8 at 56.
On agentic tasks, K3 reaches an Elo rating of 1,668 on GDPval v2, a big jump from K2.6's 1,190. It beats GLM-5.2 at 1,514, GPT-5.5 at 1,494, and Claude Opus 4.8 at 1,600, though it still falls short of Claude Fable 5 at 1,760. K3 also takes the top spot on AutomationBench-AA, Artificial Analysis's version of Zapier's agentic SaaS workflow evaluation, with a score of 53 percent.
On AA-Briefcase, a private long-horizon knowledge work evaluation, K3 reaches an overall Elo of 1,547, up 732 points from K2.6. Only Claude Fable 5 scores higher. Artificial Analysis calls K3 well-rounded, with rubric scoring and analytical quality close to Fable 5's level. GPT-5.6 Sol still leads on presentation quality, though.
K3's accuracy rate improved from 33 percent to 46 percent on the AA-Omniscience Index, pushing the overall score from +6 to +18. But its hallucination rate climbed from 39 percent to 51 percent, meaning K3 fabricates more answers even as it gets more questions right.
Designed for Full Development Projects
According to Kimi, the model's primary use case is long-running software development with minimal human oversight. K3 is built to analyze large codebases, coordinate terminal tools, and stay focused on a task across many work steps.
The model pairs programming with visual feedback. It examines screen captures, modifies code, then checks the visible output. Kimi calls this closed-loop system "Vision in the Loop" and positions it as a foundation for game development, UI design, and CAD.
Stay updated
Get the day's AI and automation news in your inbox. No spam, unsubscribe anytime.
As demos, Kimi shows off a procedurally generated 3D open-world game that K3 reportedly built entirely in the browser using Three.js, WebGPU, and GPU Compute, along with an interactive black hole visualization. For the open-world demo, K3 procedurally generated the environment and used an external tool to create the 3D rider and horse models. Other demos include a simulation of the Long March 10 rocket launch and return, plus a Game Boy Advance emulator.
Architecture and Pricing
K3 uses a mixture-of-experts architecture that activates only 16 of 896 experts at a time. It is paired with a new attention architecture called Kimi Delta Attention, which Kimi says enables up to 6.3x faster decoding for million-token contexts. "Attention residuals" reportedly boost training efficiency by about 25 percent while adding less than 2 percent in extra compute overhead.
According to the Kimi API docs, one million input tokens cost $0.30 with a cache hit and $3.00 without. One million output tokens, including reasoning, cost $15.00. These prices apply regardless of context length. Caching happens automatically, which makes unmodified long prefixes especially useful for agents and large codebases.
That puts K3 well above the price level of its predecessor K2.6, which officially costs $0.16 per million tokens with a cache hit, $0.95 without, and $4.00 for output. Chinese providers are not offering their frontier models at rock-bottom prices anymore either.
Still, K3 is much cheaper than the top Western models and sits more in the upper midrange. Anthropic's new Sonnet 5, for example, also costs $3 per million input tokens and $15 for output but delivers lower performance.
According to Artificial Analysis, K3 averages $0.94 per task on the Intelligence Index, close to GPT-5.6 Sol at $1.04 and about half the price of Opus 4.8 at $1.80. It is well above open-weight peers like GLM-5.2 at $0.32 and DeepSeek V4 Pro at $0.04, though.
K3 also uses fewer tokens than its predecessor. It needed about 132 million output tokens to complete all nine evaluations, down from roughly 166 million for K2.6, a 21 percent reduction while scoring 13 points higher. Because of the much higher token prices, K3 will likely still cost more per task than K2.6 in most cases.
Availability
K3 is already available through Kimi.com, the mobile app for iOS, Android, and HarmonyOS, the Kimi Work desktop client version 3.1.0 and later, and Kimi Code. On OpenRouter, the model is listed under the identifier "moonshotai/kimi-k3," though it is currently served there only through Moonshot itself.