$ diff k2-horizon kimi-k3
K2 Horizon vs Kimi K3
Values come from K2 Horizon and Kimi K3, where each is cited to its source. This page states no benchmark result and ranks neither model — it puts two published specifications next to each other. Where a lab has not published a figure, the row says so rather than guessing.
IfmK2 Horizon
- context
- 524K
- weights
- open
- $/M in
- —
- $/M out
- —
- context
- 1M
- weights
- open
- $/M in
- $3
- $/M out
- $15
What actually differs
- Context window
- Kimi K3 takes 1M against 524K — modestly more room in a single request.
- Weights
- Both publish weights — K2 Horizon under Apache 2.0 (weights and code), Kimi K3 under Modified MIT (commercial use permitted; weights publicly downloadable). Check the licences rather than assuming they permit the same commercial use.
- Recency
- K2 Horizon shipped 49 days after Kimi K3 (2026-09-03 vs 2026-07-16).
Full spec
| Attribute | K2 Horizon | Kimi K3 |
|---|---|---|
| Developer | Ifm | Moonshot AI |
| Released | 2026-09-03 | 2026-07-16 |
| Context window | 524,288 (flagship 375B-A23B, native from midtraining onward) | 1M tokens |
| Pricing | not recorded | $0.30/M cache-hit input · $3.00/M cache-miss input · $15.00/M output |
| License | Apache 2.0 (weights and code) | Modified MIT (commercial use permitted; weights publicly downloadable) |
| Availability | Hugging Face (weights, code, checkpoints, training data where redistributable) | Kimi Code, Kimi app, Kimi API; weights on Hugging Face (`moonshot-ai/kimi-k3`, released 2026-07-27) |