LLMsmoonshotaiVision + text4 buildsIts own licence terms685k dl/mo
Kimi-K2.6 is an open-weight language model from moonshotai. Kimi-K2.6-GGUF (GGUF) weighs 595 GB; the smallest configuration that runs it is 8× H100 80 GB.
Kimi-K2.6 is an open-source, native multimodal agentic model developed by moonshotai for image-text-to-text tasks.
It features a Mixture-of-Experts architecture with 1 trillion total parameters, 32 billion activated parameters, and a 256K context length.
The model supports image and video inputs and is released under a Modified MIT License.
Summary of the moonshotai/Kimi-K2.6 model card, 2026-10-01. The estimate below is for another build of the family.
Kimi K2.6 is an open-source, native multimodal agentic model that advances practical capabilities in long-horizon coding, coding-driven design, proactive autonomous execution, and swarm-based task orchestration.
Key Features
Long-Horizon Coding: K2.6 achieves significant improvements on complex, end-to-end coding tasks, generalizing robustly across programming languages (Rust, Go, Python) and domains spanning front-end, DevOps, and performance optimization.
Coding-Driven Design: K2.6 is capable of transforming simple prompts and visual inputs into production-ready interfaces and lightweight full-stack workflows, generating structured layouts, interactive elements, and rich animations with deliberate aesthetic precision.
Elevated Agent Swarm: Scaling horizontally to 300 sub-agents executing 4,000 coordinated steps, K2.6 can dynamically decompose tasks into parallel, domain-specialized subtasks, delivering end-to-end outputs from documents to websites to spreadsheets in a single autonomous run.
Proactive & Open Orchestration: For autonomous tasks, K2.6 demonstrates strong performance in powering persistent, 24/7 background agents that proactively manage schedules, execute code, and orchestrate cross-platform operations without human oversight.
2. Model Summary
Architecture
Mixture-of-Experts (MoE)
Total Parameters
1T
Activated Parameters
32B
Number of Layers (Dense layer included)
61
Number of Dense Layers
1
Attention Hidden Dimension
7168
MoE Hidden Dimension (per Expert)
2048
Number of Attention Heads
64
Number of Experts
384
Selected Experts per Token
8
Number of Shared Experts
1
Vocabulary Size
160K
Context Length
256K
Attention Mechanism
MLA
Activation Function
SwiGLU
Vision Encoder
MoonViT
Parameters of Vision Encoder
400M
3. Evaluation Results
General Testing Details
We report results for Kimi K2.6 and Kimi K2.5 with thinking mode enabled, Claude Opus 4.6 with max effort, GPT-5.4 with xhigh reasoning effort, and Gemini 3.1 Pro with a high thinking level.
Unless otherwise specified, all Kimi K2.6 experiments were conducted with temperature = 1.0, top-p = 1.0, and a context length of 262,144 tokens.
Benchmarks without publicly available scores were re-evaluated under the same conditions used for Kimi K2.6 and are marked with an asterisk (*). Except where noted with an asterisk, all other results are cited from official reports.
Reasoning Benchmarks
IMO-AnswerBench scores for GPT-5.4 and Claude 4.6 were obtained from z.ai/blog/glm-5.1.
Humanity's Last Exam (HLE) and other reasoning tasks were evaluated with a maximum generation length of 98,304 tokens. By default, we report results on the HLE full set. For the text-only subset, Kimi K2.6 achieves 36.4% accuracy without tools and 55.5% with tools.
Tool-Augmented / Agentic Tasks
Kimi K2.6 was equipped with search, code-interpreter, and web-browsing tools for HLE with tools, BrowseComp, DeepSearchQA, and WideSearch.
For HLE-Full with tools, the maximum generation length is 262,144 tokens with a per-step limit of 49,152 tokens. We employ a simple context management strategy: once the context window exceeds the threshold, only the most recent round of tool-related messages is retained.
For BrowseComp, we report scores obtained with context management using the same discard-all strategy as Kimi K2.5 and DeepSeek-V3.2.
For DeepSearchQA, no context management was applied to Kimi K2.6 tests, and tasks exceeding the supported context length were directly counted as failed. Scores for Claude Opus 4.6, GPT-5.4, and Gemini 3.1 Pro on DeepSearchQA are cited from the Claude Opus 4.7 System Card.
For WideSearch, we report results under the "hide tool result" context management setting. Once the context window exceeds the threshold, only the most recent round of tool-related messages is retained.
Claw Eval was conducted using version 1.1 with max-tokens-per-step = 16384.
For APEX-Agents, we evaluate 452 tasks from the public 480-task release, as done by Artificial Analysis(excluding Investment Banking Worlds 244 and 246, which have external runtime dependencies)
Coding Tasks
Terminal-Bench 2.0 scores were obtained with the default agent framework (Terminus-2) and the provided JSON parser, operating in preserve thinking mode.
For the SWE-Bench series of evaluations (including Verified, Multilingual, and Pro), we used an in-house evaluation framework adapted from SWE-agent. This framework includes a minimal set of tools—bash tool, createfile tool, insert tool, view tool, strreplace tool, and submit tool.
All reported scores for coding tasks are averaged over 10 independent runs.
Vision Benchmarks
Max-tokens = 98,304, averaged over three runs (avg@3).
Settings with Python tool use max-tokens-per-step = 65,536 and max-steps = 50 for multi-step reasoning.
MMMU-Pro follows the official protocol, preserving input order and prepending images.
4. Native INT4 Quantization
Kimi-K2.6 adopts the same native int4 quantization method as Kimi-K2-Thinking.
5. Deployment
[!Note]
You can access Kimi-K2.6's API on https://platform.moonshot.ai and we provide OpenAI/Anthropic-compatible API for you. To verify the deployment is correct, we a
Quoted from the model card on Hugging Face. The full card is behind the Hugging Face link above.