Kimi K2.6
Moonshot · Released Apr 2026 · kimi-k2.6
$0.95
Input price / 1M tokens
#23 of 61, cheapest first
$4.00
Output price / 1M tokens
#25 of 61, cheapest first
262K
Context window
#54 of 71
68
Output tokens / second
#46 of 61
80.2%
SWE-bench Verified
#2 of 12
58.6%
SWE-bench Pro
#10 of 16
66.7%
Terminal-Bench 2.0
#4 of 7
54%
Humanity’s Last Exam (with tools)
#14 of 16
Summary
Kimi K2.6 is an open-weight model from Moonshot, released on Apr 2026.
At $0.95 input and $4.00 output per million tokens, it is #23 of 61 on input price, cheapest first.
Its 262K-token context window ranks #54 of 71.
Measured output speed is 68 tokens per second, #46 of 61.
Moonshot's open multimodal agent model, 1T total and 32B active parameters, for long-horizon coding and agent swarms. Modified MIT license: commercial use allowed, but products over 100M MAU or $20M monthly revenue must display "Kimi K2.6".
Benchmarks · 6 reported
| Benchmark | Score | Bar | Rank | Source |
|---|---|---|---|---|
| SWE-bench Verified | 80.2% | #2 of 12 | Kimi-K2.6 model cardSelf-reported | |
| SWE-bench Pro | 58.6% | #10 of 16 | Kimi-K2.6 model cardSelf-reported | |
| Terminal-Bench 2.0Terminus-2 harness | 66.7% | #4 of 7 | Kimi-K2.6 model cardSelf-reported | |
| Humanity’s Last Exam (with tools)full set, with tools | 54% | #14 of 16 | Kimi-K2.6 model cardSelf-reported | |
| OSWorld-Verified | 73.1% | #8 of 8 | Kimi-K2.6 model cardSelf-reported | |
| GPQA Diamond | 90.5% | #15 of 27 | Kimi-K2.6 model cardSelf-reported |
Self-reported means the lab ran the test itself. A rank appears only where several labs report the same version of a benchmark.
Similar models
Models with the most benchmarks in common with Kimi K2.6, and the closest scores.
SWE-bench Pro
#10 of 16Harder multi-file engineering tasks
- Kimi K2.658.6
- GPT-5.558.6
- Gemini 3.1 Pro54.2
- GPT-5.457.7
- GLM-5.262.1
- Qwen3.8-Max67.7
Terminal-Bench 2.0
#4 of 7Terminal tasks, 2025 set
- Kimi K2.666.7
- GPT-5.582.7
- Gemini 3.1 Pro68.5
- GPT-5.475.1
Humanity’s Last Exam (with tools)
#14 of 16Expert questions, with search and code
- Kimi K2.654
- GPT-5.552.2
- GLM-5.254.7
- Qwen3.8-Max56.2
Compare Kimi K2.6 side by side
Pre-filled with the two closest models. Swap any of them.
Sources
- price, context: platform.moonshot.ai/docs/pricing/chat, checked 2 Oct 2026
- release month (April 2026): platform.kimi.ai/docs/platform-changelog, checked 2 Oct 2026
- modalities: platform.moonshot.ai/docs/guide/kimi-k2-6-quickstart, checked 2 Oct 2026
- parameters, license, benchmarks: huggingface.co/moonshotai/Kimi-K2.6, checked 2 Oct 2026
- output speed: artificialanalysis.ai/models/kimi-k2-6, checked 2 Oct 2026
Building on Kimi K2.6?
Get the architecture right before the bill arrives.
We size caching, routing and fallbacks for your workload on a free call, and tell you where a cheaper model is good enough.