Moonshot AI Releases Kimi K3: Open-Weight Multimodal Reasoning Model with 1M Context Window
Moonshot AI has released Kimi K3, an open-weight multimodal reasoning model with a 1-million token context window. The model is priced at $3 per 1M input tokens and $15 per 1M output tokens, available through OpenRouter.
Kimi K3 — Quick Specs
Moonshot AI Releases Kimi K3: Open-Weight Multimodal Reasoning Model with 1M Context Window
Moonshot AI has released Kimi K3, an open-weight multimodal reasoning model with a 1-million token context window. The model is priced at $3 per 1M input tokens and $15 per 1M output tokens, available through OpenRouter.
Model Specifications
Kimi K3 is described by Moonshot AI as an "ultra-large-scale" model designed for complex coding, knowledge work, and long-horizon agentic workflows. The model supports both input and output modalities, with a context window of 1 million tokens — matching the largest available context windows in current AI models.
The model was released on July 15, 2026, according to OpenRouter's listing. It uses Key-Dense Attention (KDA) and Attention Residuals architecture for computational efficiency, though specific parameter counts have not been disclosed.
Capabilities and Use Cases
According to Moonshot AI, Kimi K3 is particularly suited for:
- Navigating large code repositories
- Tool use and debugging
- Iterating against images, logs, tests, and runtime feedback
- Complex coding tasks
- Knowledge work requiring extended context
The model's 1M token context window enables it to process entire codebases or lengthy documents in a single request, which is significant for developers working with large-scale projects.
Availability and Pricing
Kimi K3 is available exclusively through OpenRouter, which forwards requests directly to the model without routing decisions. The pricing structure is:
- Input: $3 per 1M tokens
- Output: $15 per 1M tokens
OpenRouter notes that actual costs can be 60-80% lower than list prices when prompt caching is utilized for repeated context.
What This Means
Kimi K3's release adds another option to the growing field of large-context models, competing with offerings like Anthropic's Claude 3.5 Sonnet (200K context) and Google's Gemini 1.5 Pro (2M context). The "open-weight" designation suggests the model weights are available for download, though licensing terms have not been specified. At $3/$15 per 1M tokens, Kimi K3's pricing is competitive with mid-tier models, though significantly more expensive than some open-source alternatives. The model's focus on coding and agentic workflows positions it as a developer-focused tool, particularly for teams working with large codebases that require multimodal reasoning capabilities.
Related Articles
Unbiased releases Pareto 26.9: 262K-context composite model at $2.50/$7.50 per 1M tokens
Unbiased has released Pareto 26.9, a multimodal composite model with a 262K-token context window, priced at $2.50 per 1M input tokens and $7.50 per 1M output tokens. It is listed on OpenRouter with a Oct 13, 2026 release date. A cheaper 1.0M-context preview, Pareto 26.10, is already listed alongside it.
Liquid AI releases LFM2.5-24B-A2B: open-weight 24B reasoning MoE with 2B active parameters and 131K context
Liquid AI released LFM2.5-24B-A2B, an open-weight reasoning mixture-of-experts model with 24B total and 2B active parameters and a 131K-token context window. The company claims it is competitive with Qwen3.6-35B-A3B on agentic and tool-use tasks while decoding at 184 tokens/s on an Apple M5 Max.
ScaleDown launches Summarize, a 1M-context abstractive summarization model at $0.04 per 1M input tokens on OpenRouter
ScaleDown has released Summarize, an abstractive summarization model with a 1M-token context window, listed on OpenRouter on October 13, 2026. OpenRouter shows $0.04 per 1M input tokens (20% off the $0.05 list price) and $0 per 1M output tokens. Parameter count and benchmark scores have not been disclosed.
OpenBMB releases MiniCPM-V 4.7 1B, an on-device vision model with 4x/16x visual token compression
OpenBMB has released MiniCPM-V 4.7 1B, a 1B-parameter multimodal model for single-image, multi-image and video understanding on phones, tablets and local hardware. OpenBMB claims it substantially outperforms Qwen3.5 0.8B in overall score and inference efficiency in Instruct mode. Benchmark figures, context window and training cutoff are not given in the available model card text.
Comments
Loading...