Moonshot: Kimi K3.1 Pro
Multimodal Kimi model with 1M context and always-on reasoning for coding, 3D creation, financial research, and agentic work
Sources
Same model, different sources. Our first-party rate is listed first, then 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 |
|---|---|---|---|---|---|
| tokenstatmoonshotai/kimi-k3.1-pro | Moonshot AI | $2 | $10 | $0.2 | 1.0M |
| models.devcheapestkimi-k3.1-pro | Moonshot AI (China) | $1.94 | $10.17 | $0.1944 | 1.0M |
| models.devmoonshot/kimi-k3.1-pro | Eden AI | $2 | $10 | $0.2 | 1.0M |
| models.devmoonshot/kimi-k3.1-pro | LLM Gateway | $2 | $10 | $0.2 | 1.0M |
| models.devmoonshot/kimi-k3.1-pro | Merge Gateway | $2 | $10 | $0.2 | 1.0M |
| models.devkimi-k3.1-pro | DevPass (LLM Gateway) | $2 | $10 | $0.2 | 1.0M |
| models.devkimi-k3.1-pro | OpenCode Go | $2 | $10 | $0.2 | 1.0M |
| models.devkimi-k3.1-pro | Requesty | $2 | $10 | $0.2 | 1.0M |
| models.devmoonshotai/kimi-k3.1-pro | Kilo Gateway | $2 | $10 | $0.2 | 1.0M |
| models.devkimi-k3.1-pro | Moonshot AI | $2 | $10 | $0.2 | 1.0M |
| models.devmoonshotai/kimi-k3.1-pro | NanoGPT | $2 | $10 | $0.2 | 1.0M |
| models.devmoonshotai/kimi-k3.1-pro | NEAR AI Cloud | $2 | $10 | $0.2 | 1.0M |
| models.devmoonshotai/kimi-k3.1-pro | Novita AI | $2 | $10 | $0.2 | 1.0M |
| OpenRoutermoonshotai/kimi-k3.1-pro | OpenRouter | $2 | $10 | $0.2 | 1.0M |
| models.devmoonshotai/kimi-k3.1-pro | Prime Inference | $2 | $10 | $0 | 1.0M |
| models.devmoonshotai/kimi-k3.1-pro | Vercel AI Gateway | $2 | $10 | $0.2 | 1.0M |
| models.devmoonshotai/kimi-k3.1-pro | Ofox | $2 | $11 | $0.2 | 1.0M |
| LiteLLMmoonshot/kimi-k3.1-pro | N/A | $2 | $10 | $0.2 | N/A |
| LiteLLMkimi-k3-1-pro | N/A | $2 | $10 | $0.2 | N/A |
| models.devkimi-k3.1-pro | 302.AI | $2.40 | $12 | $0.24 | 1.0M |
| LiteLLMkyma/kimi-k3.1-pro | N/A | $2.70 | $13.50 | $0.27 | N/A |
| models.devkimi-k3-1-pro | Venice AI | $3.75 | $18.75 | $0.375 | 1M |
Capabilities
- Reasoning
- Tool calling
- Structured output
- Attachments
- In: text
- In: image
- In: video
- Out: text
Scores
Vendor launch report. Launch results use max reasoning effort, temperature 1.0 and top-p 0.95 unless stated otherwise. Coding uses Claude Code; other evaluations use benchmark-specific harnesses. BenchCAD uses a 1,000-case subset. PostTrainBench v1.2 uses H20 GPUs and a DeepSeek judge. Agents’ Last Exam v1.1 rejects runs that retrieve answers instead of completing the task. Internal tests and tool-assisted scores are labelled separately.
Sparse coverage on purpose. Missing a score means we have not linked one yet, not that the model is unranked. Hover or tap the i next to any score for its source. Full boards on /benchmarks.
- Terminal-Bench 4.065.7i
- Terminal-Bench Science 0.138.1i
- DeepSWE v1.175.5i
- APEX-Agents67.2i
- AA-Briefcase1678i
- HLE (without tools)51.1i
- HLE (with tools)70i
- GDPVal-AA v2.11846i
- GPQA Diamond94.2i
- ExploitBench74.7i
- Chartography (with tools)90.2i
- GDP.pdf34.6i
- ProgramBench (at least 95% tests)31.5i
- SWE-Marathon v1.151.7i
- BenchCAD (1,000 cases, no tools)72.8i
- BenchCAD (1,000 cases, with tools)92.7i
- PostTrainBench v1.2 (H20, DeepSeek judge)36.5i
- Kimi Code Bench 2.0 (internal)77.1i
- Kimi Code Bench 3.0 Long-horizon (internal)61.2i
- BrowseComp92.1i
- BioMysteryBench v1180.6i
- CharXiv RQ (with Python)93.4i
- MMMU-Pro (with Python)90i
- GDP.pdf (with Python, rubric pass rate)85.7i
- BabyVision (with Python)95.4i
- ZeroBench main (with Python, pass@5)63i
- Agents’ Last Exam v1.1 (all, Kimi protocol)43.2i
- Agents’ Last Exam v1.1 (CLI, Kimi protocol)44.2i
- AutomationBench-AA (vendor-cited)73.3i
- OfficeQA Pro v264.4i
- OfficeQA Pro67.4i
- Frontier Finance58.6i
- Harvey Legal Agent (all-pass rate)23.7i
- HANDBOOK.md v144.6i
- OSWorld 2.1 (Kimi Agent)79.6i
- DECK-Arena (internal Elo)1456i
- Design Bench (internal Elo)957i
- Kimi Hallu Bench Hard (tools, lower is better)8.3i
- HLE-Diamond39.4i
- CharXiv RQ (no tools)89i
- MMMU-Pro (no tools)84.1i
- ZeroBench main (no tools, avg@5)29.8i
- ZeroBench main (no tools, pass@5)39i
- PerceptionBench68.3i
- Kimi Hallu Bench (no tools, lower is better)5.4i
- ExploitGym (6-hour budget)20.7i
- xdev-bench-hard (internal, pass@3)81i
- Financial Modeling (Kimi internal)57.6i
- Financial Research (Kimi internal)70.4i
Arena rows are community-vote Elo from Arena AI, Artificial Analysis indices arrive via OpenRouter, and launch-eval figures are collected from vendor announcements, analyst writeups, and community benchmarks. Full boards and credits on /benchmarks.
See what this model costs you
The app reads the sessions your tools already wrote and prices this model's tokens at the published API rate. Free on your computer.