Kimi K2 Thinking
Best for beginners. Describe what you want in a chat and the picture comes back, no learning curve.
Images and video from a chat, with short commands. The most convenient tool for Telegram users, with a web companion.
Best for power users. Full control over models, settings and workflows in a browser studio.
Best for communities. Add it to your own server and run any workflow from a channel.
One screen for every model and workflow: prompts, references, edits and video, built into your Graydient account.
Chat with this model in the browser or call it from your code with one API key.
Set up local programs if your computer has capable hardware.
Generate the same prompt on up to 50 checkpoint/base models at a time.
MoonshotAI | 262,144 token context | $0.60/M input, $2.50/M output. Kimi K2 Thinking is Moonshot AI’s most advanced open reasoning model to date, extending the K2 series into agentic, long-horizon reasoning. Built on the trillion-parameter Mixture-of-Experts (MoE) architecture introduced in Kimi K2, it activates 32 billion parameters per forward pass and supports 256 k-token context windows. The model is optimized for persistent step-by-step thought, dynamic tool invocation, and complex reasoning workflows that span hundreds of turns. It interleaves step-by-step reasoning with tool use,…MoonshotAI | 262,144 token context | $0.60/M input, $2.50/M output. Kimi K2 Thinking is Moonshot AI’s most advanced open reasoning model to date, extending the K2 series into agentic, long-horizon reasoning. Built on the trillion-parameter Mixture-of-Experts (MoE) architecture introduced in Kimi K2, it activates 32 billion parameters per forward pass and supports 256 k-token context windows. The model is optimized for persistent step-by-step thought, dynamic tool invocation, and complex reasoning workflows that span hundreds of turns. It interleaves step-by-step reasoning with tool use, enabling autonomous research, coding, and writing that can persist for hundreds of sequential actions without drift. It sets new open-source benchmarks on HLE, BrowseComp, SWE-Multilingual, and LiveCodeBench, while maintaining stable multi-agent behavior through 200–300 tool calls. Built on a large-scale MoE architecture with MuonClip optimization, it combines strong reasoning depth with high inference efficiency for demanding agentic and analytical tasks.
Captain's notes
Known for
- Maintains coherent tool-use across 200–300 sequential calls without drift.
- Native INT4 quantization-aware training delivers roughly 2x inference speed-up.
- 256k-token context window for long documents and extended agentic sessions.
- Top-tier open-source scores on HLE, AIME25, and LiveCodeBench.
- Interleaves chain-of-thought reasoning with live tool invocation end-to-end.
Watch out for
- At 1 trillion total parameters, self-hosting demands serious GPU infrastructure despite INT4 efficiency.
- Output pricing at $2.50/M tokens adds up fast in long agentic runs with hundreds of tool calls.
- Moonshot AI has not published the licence terms openly; check the repository before any commercial use.
Prompts to try
Compare with
- Kimi K3Open-weights model1,048,576 token context · $0.30/M input, $11.94/M output
- Kimi K2.7 CodeOpen-weights model262,144 token context · $0.67/M input, $3.35/M output
- Kimi K2.5Open-weights model262,144 token context · $0.50/M input, $2.50/M output
- Kimi K2 0711Open-weights model131,072 token context · $0.57/M input, $2.30/M output
Questions people ask
- Can I run Kimi K2 Thinking locally?
- Yes, the weights are open and available on Hugging Face under moonshotai/Kimi-K2-Thinking. The model uses native INT4 quantization to reduce memory and latency, but at 1 trillion total parameters you will still need substantial GPU hardware.
- What does Kimi K2 Thinking cost via API?
- Moonshot AI prices it at $0.60 per million input tokens and $2.50 per million output tokens, with a 262,144-token context window.
- How does Kimi K2 Thinking differ from the base Kimi K2?
- K2 Thinking extends the base K2 with deep chain-of-thought reasoning and end-to-end tool orchestration, allowing it to sustain coherent agentic workflows across hundreds of steps where the base model is not optimized for that kind of long-horizon behavior.
Captain's verdictIf you need an open-weights reasoning agent that actually holds together across a long, tool-heavy workflow, K2 Thinking is the most credible option available right now.
Glossary: the terms used here
- Heat score
- One number for how much a model is used and wanted right now, from 0° to 9,999°. Every list on this site that says hottest is sorted by it and nothing else.
- Floor
- The 7,000° every model with real data starts from. The scale runs from 0° to 9,999°; the part below the floor is reserved for stand-ins (models without data), so that a model with data always ranks above one without. The floor is the same for everyone and decides nothing; the degrees above it are the measurement.
- Download-equivalents
- Everything the score treats as a download, added up: downloads themselves, each torrent seeder as 1,000, each AI Horde image and each run on a generation site as one.
- Log scale
- Counting by tens, not by ones: going from 1,000 to 10,000 downloads adds the same degrees as going from 10,000 to 100,000. It keeps giants from crowding out everything else.
- Likes
- Thumbs up and likes from every source, worth half a download each on the log scale.
- Adoption rate
- How many downloads a day a model gets right now, from the difference between two readings of a source. A quarter of the weight of the download term.
- Measured and estimated
- Measured means two readings of a source exist and the difference is real. Estimated means only today's total is known and the past is spread over the model's life; the site says which one it is showing.
- Seeder
- Someone who keeps a copy of the weights online for others to download over a torrent, or a hosted mirror doing the same. Each one counts as 1,000 downloads.
- AI Horde
- A volunteer network that generates images for free on donated computers. The number of images it has made with a model is public, and each counts as one download.
- Generation sites
- Sites that host their own copies of a model and let people generate with it, such as PixAI, SeaArt and Yodayo. Their downloads count like Civitai's and each generation run as one download.
- OpenRouter tokens
- A text model served through an API, such as Claude or GPT, has no weights to download, so its usage is measured in tokens: the pieces of text it read and wrote through OpenRouter, a gateway many apps use, over the last 30 days. Every 10,000,000 tokens count as one download. Source: OpenRouter (openrouter.ai/rankings), CC BY 4.0.
- Full model (checkpoint)
- The weights themselves, the file that generates on its own. It gets a bonus because add-ons depend on it.
- LoRA and embedding
- Small add-on files that steer someone else's weights toward a style, a character or a concept. They cannot run alone.
- Family and weights
- The base a model is built on, such as SD 1.5, SDXL, Flux or Wan. Every model inherits the release date of its family's weights.
- Age
- Degrees taken off for how long ago the family's weights came out: 150° a year, up to 750°. It keeps old downloads that had years to pile up from freezing the top of the list.
- Family boost
- A small, bounded nudge for LoRAs of a family this site is pushing forward right now, always listed in the breakdown.
- Stand-in score
- A stable number below the floor given to a model no source reports on yet, so lists stay complete. It never ranks above a model with data.
- New! (grace period)
- A model less than 7 days old shows New! instead of a score and holds no place in the rankings. Its score is computed from day one, but a model one day old grows faster than anything, so without the wait the week's #1 would always be whatever came out yesterday rather than what held up for a week.
- Trend
- How a model's growth over a period compares with the typical model's growth over the same period. Faster than typical reads as a rise, slower as a fall.
- Typical model
- The middle of the catalog: half the models with a figure grew faster over the period, half slower.
