Kimi K3
Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work
Cheapest input
$2.70
$/MTok · models.dev
Cheapest output
$13.50
$/MTok
Context
1.0M
tokens
Offers
17
2 sources
Offers
Same model, different listings. First-party, router, and aggregator rows stay separate so you can see the spread instead of a single blended number.
| Source | Provider | Input | Output | Cache read | Context |
|---|---|---|---|---|---|
| models.devcheapestmoonshotai/Kimi-K3 | DeepInfra | $2.70 | $13.50 | $0.27 | 1.0M |
| models.devmoonshotai/kimi-k3 | Ai& | $3 | $12.50 | $0.5 | 1.0M |
| models.devmoonshotai/Kimi-K3 | Baseten | $3 | $15 | $0 | 1.0M |
| models.devmoonshotai/Kimi-K3 | Berget.AI | $3 | $15 | $0 | 328K |
| models.devmoonshotai/kimi-k3 | Cloudflare | $3 | $15 | $0.3 | 1.0M |
| models.devmoonshotai/Kimi-K3 | Hugging Face | $3 | $15 | $0 | 1M |
| models.devmoonshotai/kimi-k3 | Impossibl | $3 | $15 | $0.3 | 1.0M |
| models.devmoonshotai/kimi-k3 | Kilo Gateway | $3 | $15 | $0.3 | 1.0M |
| models.devmoonshotai/Kimi-K3 | Modal | $3 | $15 | $0.3 | 1.0M |
| models.devkimi-k3 | Moonshot AI | $3 | $15 | $0.3 | 1.0M |
| models.devmoonshotai/Kimi-K3 | Nebius | $3 | $15 | $3 | 1.0M |
| models.devmoonshotai/kimi-k3 | Novita AI | $3 | $15 | $0.3 | 1.0M |
| OpenRoutermoonshotai/kimi-k3 | OpenRouter | $3 | $15 | $0.3 | 1.0M |
| models.devmoonshotai/Kimi-K3 | Together AI | $3 | $15 | $0.3 | 1.0M |
| models.devmoonshotai/kimi-k3 | Vercel AI Gateway | $3 | $15 | $0.3 | 1M |
| models.devmoonshotai/Kimi-K3 | Weights & Biases | $3 | $15 | $0.3 | 1.0M |
| models.devmoonshotai/kimi-k3 | ZenMux | $3 | $15 | $0.3 | 1.0M |
Capabilities
- Reasoning
- Tool calling
- Structured output
- Attachments
- Open weights
- In: text
- In: image
- In: pdf
- In: video
- Out: text
Released: 2026-07-16Family: kimi-k3
Scores
Sparse coverage on purpose. Missing a score means we have not linked one yet, not that the model is unranked. Full boards on /benchmarks.
- Arena agent (net improvement)9.91
- Arena agent net improvement9.91
- Arena agent confirmed success14.23
- AA intelligence57.1
- AA coding76.2
- AA agentic50.1
Sources: Arena AI, Artificial Analysis. See /benchmarks.