- Published on
Tencent WorkBuddy Integrates Kimi-K3.1-Pro Model, Supports 1 Million Context
- Authors
- Name
- DeepHorus
On October 24, Tencent WorkBuddy announced that it has officially integrated the Kimi-K3.1-Pro model and simultaneously launched the high-speed version Kimi K3.1 Pro Highspeed. According to the introduction, while maintaining the same level of intelligence, the high-speed version's output speed is approximately 2 to 3 times that of the regular version. Tencent stated that due to current resource constraints, the model is currently prioritized only for the "Enterprise Flagship Edition" and "individual subscribers."
Public information shows that Kimi K3.1 Pro is the first model released in the new K3.1 series and is also Kimi's most capable 3-trillion-parameter model to date, supporting a 1-million-token context window. During R&D, the team substantially increased compute investment in the reinforcement learning (RL) stage, significantly enhancing the model's reasoning and tool-calling capabilities.
In terms of capability distribution, K3.1 Pro shows particularly significant improvements in coding, 3D creation, financial research, and other long-horizon Agent tasks. According to the official statement, on coding, Agent, and knowledge benchmarks, the model has reached or approached the level of the strongest closed-source models, which has also become an important basis for its early adoption by office collaboration platforms.
In the Terminal-Bench 4.0 evaluation, which examines the real-environment task capabilities of coding agents, K3.1 Pro scored 65.7%, surpassing GPT-6 Astra's 58.2% and only slightly trailing Claude Opus 5.5's 66.4%. This benchmark tests item by item through resource-constrained terminal tasks, with each task having strict specialized assessments.
3D creation and spatial reasoning are another strong area of K3.1 Pro. In the BenchCAD benchmark, which reconstructs 3D models by generating CAD code from multi-view rendered images, K3.1 Pro achieved a score of 92.7% with the help of tools, close to Claude Opus 5.5's 96.2% and GPT-6 Astra's 95.9%.
With stronger reasoning and tool-calling capabilities, K3.1 Pro can deliver complete financial workflows end to end. In external benchmarks, the model scored 64.4 and 36.3 on OfficeQA Pro v2 and SpreadsheetBench v2, respectively, both setting new current state-of-the-art (SoTA) results.
This integration means WorkBuddy users can directly invoke million-token context and long-horizon Agent capabilities in office scenarios to handle composite tasks such as code writing, document analysis, and spreadsheet calculations. Against the backdrop of domestic large models accelerating their embedding into enterprise office tools, the combination of model capabilities and platform entry points is becoming a new focus of competition, and the subsequent scope of availability and quota arrangements remain to be seen.
