Kimi K3
by Charm Hyper · family kimi-k3 · listed Jul 2026RTVW
Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work
Current rates USD per 1M tokens
Input$3.27
Output$16.33
Cache read$0.327
Cache write—
Against the market
On the record
- Model id
kimi-k3- Context window
- 1.0M tokens
- Max output
- 16K tokens
- Input modalities
- text, image
- Output modalities
- text
- Reasoning
- yes
- Tool calls
- yes
- Open weights
- yes
- Knowledge cutoff
- —
- Released
- 2026-07-16
- Last updated
- 2026-07-16
- Provider docs
- hyper.charm.land
Back of the envelope
≈
Also listed at
- SenseNova (China)free in · free out
- Volcengine Ark Coding Planfree in · free out
- SCNet Token Planfree in · free out
- Nvidiafree in · free out
- OpenRouter$0.159 in · $20.00 out
- Kenari$0.857 in · $8.57 out
- NanoGPT$1.65 in · $8.25 out
- engy$1.95 in · $9.75 out
- ainetcafe$2.10 in · $10.50 out
- Boundless$2.30 in · $11.40 out
- IteraCompute$2.35 in · $11.76 out
- Vancine$2.40 in · $12.00 out
- SiliconFlow$2.70 in · $13.50 out
- Wallaby$2.70 in · $13.50 out
- Alibaba (China)$2.83 in · $14.13 out