See key specs and per-benchmark scores for each model/mode. Scroll horizontally for all columns. 当前对比 2 个模型的评测数据与核心参数。
Inkling
Thinking Machines Lab
Each axis is a category average, normalized to a 100-point radar.
Relative edge: none clear / Relative gap: AI Agent - 工具使用 -16.8
Relative edge: AI Agent - 工具使用 +16.8 / Relative gap: none clear
Method: for each model and benchmark, the chart first averages all scores in the current mode scope instead of taking the best score, then averages those benchmark scores within each category. Only benchmarks with at least two selected models scored are included; missing values are not counted as zero.
Best overall
Kimi K3 · 82.64
Best single
Kimi K3 · GPQA Diamond 93.50
Modality coverage
Inkling · 3 modalities
Head to head
5
Benchmarks
0
Wins
5
Losses
-12.62
Average diff
Compare benchmark results across thinking modes and tool usage.
Data sourced primarily from official releases (GitHub, Hugging Face, papers), then benchmark leaderboards, then third-party evaluators. Learn about our data methodology
Complete scores for each model/mode across selected benchmarks.
5 benchmarks with comparable scores. Each model shows its best score; mode label is displayed below.
| Benchmark | Inkling | Kimi K3 |
|---|---|---|
HLE 综合评估 | 46.00Thinking Enabled ｜ Tools | 56.00Thinking Level · High ｜ Tools |
GPQA Diamond 科学与综合推理 | 87.20Thinking Enabled | 93.50Thinking Level · High |
BrowseComp AI Agent - 信息收集 | 77.10Thinking Enabled ｜ Tools | 91.20Thinking Level · High ｜ Tools |
MCP-Atlas AI Agent - 工具使用 | 76.00Thinking Level · Extra High ｜ Tools | 84.20Thinking Level · High ｜ Tools |
Terminal-Bench 2.1 AI Agent - 工具使用 | 63.80Thinking Enabled ｜ Tools | 88.30Thinking Level · High ｜ Tools |
Side-by-side input/output token pricing
Licensing, MoE architecture, and multi-modality support.
| Features & specs | InklingThinking Machines Lab | Kimi K3Moonshot AI |
|---|---|---|
Core specsRelease | 2026-07-15 | 2026-07-16 |
Context length | 1M | 1M |
Parameters | 975B | 2.8T |
Active parameters | 41B | 104B |
Max output | Not provided | 1048576 |
MoE | Yes | Yes |
Supported modes | 常规模式（Non-Thinking Mode）可控思考强度（effort 0.2-0.99） | No mode data |
LicenseCode Open Source | Open Source · Apache 2.0 | Open Source · Kimi K3 License |
Weights Open Source | Open Source · Apache 2.0 | Open Source · Kimi K3 License |
Commercial use | 免费商用授权 | 有条件免费商用授权 |
Local deploymentWeight size | Not provided | 约 1.42 TiB（MXFP4，96 个 Safetensors 分片） |
VRAM for weights | Not provided | ≈ 1454 GB (weights only, excludes KV cache) |
Weights | Hugging Face | Hugging Face |
Source repo | Not provided | GitHub |
Modality supportText Input/Output | / | / |
Image Input/Output | / | / |
Audio Input/Output | / | Not provided |
Video Input/Output | Not provided | / |
ResourcesPaper / report | Not provided | Kimi K3 技术报告 |
Kimi K3
Moonshot AI
