MoonshotAI: Kimi K2.6
Moonshot
Capabilities
Strengths
- 262K context window enables very long document processing
- Strong multimodal capabilities (vision + text)
- Open weights allow self-hosting and full data privacy
- Extended thinking for complex reasoning tasks
- Excellent long-horizon coding performance
Weaknesses
- Extremely slow inference (0.03 speed score, ~33 tok/s)
- Output pricing $2.45/M tokens is mid-range cost
- Limited provider transparency on training data
- Vision capability not fine-tuned as deeply as flagship vision models
- No known sustainability certifications or green infrastructure
Pricing
Input / 1M tokens
$0.4375
Output / 1M tokens
$2.4499999999999997
Context window
262.144k
Transparency
Open weights
10.0 / 10
Open training data
0.0 / 10
Open methodology
7.0 / 10
Licence openness
9.0 / 10
Provider disclosure
6.0 / 10
FMTI company score
N/A
Kimi K2.6 ships with open weights, enabling self-hosting and full data control. Moonshot publishes model cards and some architectural details, but training data composition and full methodology remain proprietary. Open-weight status elevates privacy potential significantly.
Sustainability
Inference energy
N/A
Training footprint
N/A
Provider infrastructure
5.0 / 10
No public sustainability commitments or energy/carbon reporting from Moonshot. Large model size (likely 100B+ parameters) implies substantial inference costs. Provider infrastructure sustainability unknown.