Model
Kimi K3
Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work
12 providersreleased open weightsbeta
$2.50
best input /M · NanoGPT
$13.50
best output /M
1.05M
context window
131K
max output
Compare providers (sorted by listed input price)
| Provider | Input /M | Output /M | Cache read /M | Cache write /M | Context | Capabilities | Status | Updated |
|---|---|---|---|---|---|---|---|---|
| NanoGPT moonshotai/kimi-k3 | $2.50 | $13.50 | $0.25 | — | 1.05M | reasoningtoolsstructuredvision | Jul 16, 2026 | |
| Impossibl moonshotai/kimi-k3 | $3 | $15 | $0.3 | — | 1.05M | reasoningtoolsstructuredvision | Jul 16, 2026 | |
| OpenRouter moonshotai/kimi-k3 | $3 | $15 | $0.3 | — | 1.05M | reasoningtoolsstructuredvision | Jul 16, 2026 | |
| Moonshot AI kimi-k3 | $3 | $15 | $0.3 | — | 1.05M | reasoningtoolsstructuredvision | Jul 16, 2026 | |
| ZenMux moonshotai/kimi-k3 | $3 | $15 | $0.3 | — | 1.05M | reasoningtoolsstructuredvision | Jul 16, 2026 | |
| Arcee moonshotai/kimi-k3 | $3 | $15 | $0.3 | — | 1.05M | reasoningtoolsstructuredno vision | beta | Jul 16, 2026 |
| Ofox moonshotai/kimi-k3 | $3 | $15 | $0.3 | — | 1.05M | reasoningtoolsstructuredvision | Jul 16, 2026 | |
| Vercel AI Gateway moonshotai/kimi-k3 | $3 | $15 | $0.3 | — | 1M | reasoningtoolsstructuredvision | Jul 16, 2026 | |
| NovitaAI moonshotai/kimi-k3 | $3 | $15 | $0.3 | — | 1.05M | reasoningtoolsstructuredvision | Jul 16, 2026 | |
| ai& moonshotai/kimi-k3 | $3 | $12.50 | $0.5 | — | 1.05M | reasoningtoolsstructuredvision | Jul 16, 2026 | |
| TensorX moonshotai/kimi-k3 | $3 | $15 | $0.75 | — | 1.05M | reasoningtoolsstructuredvision | Jul 16, 2026 | |
| Kilo Gateway moonshotai/kimi-k3 | $3 | $15 | $0.3 | — | 1.05M | reasoningtoolsstructuredvision | Jul 16, 2026 |
Price badge
https://models.sutraworks.ai/badge/moonshotai/kimi-k3.svgLive SVG, regenerated on every hourly sync — paste it into a README to always show the current cheapest listed price. Updates when prices change; no build step needed on your side.
Benchmarks
- DeepSWE(resolve rate)report ↗67.5
- Terminal-Bench ↗(accuracy)report ↗88.3
- FrontierSWE ↗(dominance score)report ↗81.2
- Program Bench(score)report ↗77.8
- SWE Marathon ↗(resolve rate)report ↗42
- GDPval-AA(Elo)report ↗1668
- AA-Briefcase(Elo)report ↗1548
- AutomationBench(success rate)report ↗30.8
- JobBench(score)report ↗52.9
- SpreadsheetBench(score)report ↗34.8
- BrowseComp ↗(accuracy)report ↗91.2
- CharXiv Reasoning(accuracy)report ↗91.3
- ZeroBench(pass@5)report ↗41
Prices are per million tokens (USD). “—” means the provider does not publicly list a price for this model. Data from models.dev; verify with the provider before purchasing.