English | 中文
A lightweight preview model taking a major step toward Delegated Intelligence.
SenseNova 6.8 Flash Lite Preview is now available through the SenseNova Token Plan. It can autonomously plan, execute over long horizons, coordinate multiple agents, correct its course, and deliver end-to-end results — moving beyond answering questions toward completing delegated work.
Important
This is an early preview release. Model behavior, availability, and interfaces may evolve before the official release. SenseNova 6.8 Flash Lite and the more powerful SenseNova 6.8 Flash are coming soon.
SenseNova 6.8 Flash Lite Preview strengthens the three capabilities required to turn long, complex requests into completed deliverables:
- Maintains goals, constraints, and critical facts across hundreds of steps and multiple phases
- Keeps working through tasks that can span hours
- Detects failures, rolls back when needed, and replans toward the original objective
- Lets a lead agent dynamically organize and dispatch more than ten specialist agents
- Runs research, data analysis, mathematical reasoning, visual understanding, and fact-checking in parallel
- Reconciles intermediate findings into a single coherent result
- Understands text, images, charts, documents, video, web pages, and application interfaces
- Plans, reasons, calls tools, and verifies results along one unified task trajectory
- Connects multimodal understanding directly to actions and final deliverables
In a representative workflow, the model can independently review and analyze news, policies, market data, and other multimodal signals; identify key variables; reason about how events may evolve; and produce an in-depth analysis and forecast report for topics such as market dynamics, sporting events, and emerging situations.
Define the objective
↓
Collect multimodal evidence
↓
Dispatch specialist sub-agents
↓
Analyze, cross-check, and replan
↓
Build scenarios and forecasts
↓
Verify and deliver the final report
This is not merely information retrieval. It is a complete loop of plan → act → coordinate → correct → verify → deliver.
Public demo materials and complete evaluation results will be added as they become available.
- Dynamic presentations — research and write the content, create charts, design the layout, and deliver native HTML presentations that go beyond static slides
- Browser and app operation — understand multimodal web pages and application interfaces, then operate them to complete office workflows
- Local-file workflows — read and organize local documents, connect information across files, and produce finished artifacts
- Coding and automation — write scripts, run automated workflows, inspect the outcome, and iterate when execution fails
End-to-end workflows require an agent runtime, appropriate tools, and Skills. A direct model API call alone does not provide browser or application control.
-
Register and complete identity verification at SenseNova Console.
-
From the console sidebar, go to Management Center → API Key Management → Create API Key, then copy and store it safely. The full key is shown only once.
-
Set the environment variable:
export SENSENOVA_API_KEY="your_api_key_here"
curl 'https://token.sensenova.cn/v1/chat/completions' \
-H "Authorization: Bearer $SENSENOVA_API_KEY" \
-H 'Content-Type: application/json' \
-d '{
"model": "sensenova-6.8-flash-lite",
"max_tokens": 2000,
"messages": [{"role": "user", "content": "Plan and summarize the key steps for a competitive analysis."}]
}'For multi-turn conversations, multimodal input, streaming, the OpenAI SDK, and error handling, see API.md.
To run end-to-end delegated workflows, pair SenseNova 6.8 Flash Lite Preview with an agent runtime and the official skill library.
- Recommended runtime: OpenClaw or hermes-agent
- Model API: SenseNova Token Plan
- Skills: OpenSenseNova/SenseNova-Skills
- Installation guide: SenseNova-Skills INSTALL.md
During the preview period, the shared installation guide may still show the 6.7 model name. When configuring your provider for this preview, use
sensenova-6.8-flash-lite.
The simplest route is to ask your agent to install the repository:
"Please install https://github.com/OpenSenseNova/SenseNova-Skills into your skills directory."
After installation, you may need to restart the agent service before the new skills are discovered.
| Agent | Target directory |
|---|---|
| OpenClaw | ~/.openclaw/skills/ |
| hermes-agent | ~/.hermes/skills/ |
Prefer to install manually?
git clone https://github.com/OpenSenseNova/SenseNova-Skills.git --depth=1
mkdir -p ~/.openclaw/skills
cp -r SenseNova-Skills/skills/* ~/.openclaw/skills/For Hermes, replace the target directory with ~/.hermes/skills/.
| Stage | Model | Status |
|---|---|---|
| Early preview | SenseNova 6.8 Flash Lite Preview | Available now on Token Plan |
| Official lightweight release | SenseNova 6.8 Flash Lite | Coming soon |
| More powerful release | SenseNova 6.8 Flash | Coming soon |
| Evaluations and cases | Full benchmark results and application showcases | Planned for August |
SenseNova 6.8 Flash Lite Preview is available now through the SenseNova Token Plan, with a limited-time free offer for early users.
Have integration questions, want to share a workflow, or found a preview issue? Join the community to reach the team, exchange ideas, and help shape the official release.
| Discord | WeChat Group | Feishu Group |
 |
 |
 |
- Website: sensenova.cn
- Experience platform: SenseNova Token Plan
- API documentation: API.md
- Official Skills: OpenSenseNova/SenseNova-Skills