GLM-5.2 Free Guide: SenseTime SenseNova Token Plan, Flagship Model for Free

There is a strange phenomenon in the domestic AI circle: the stronger the model, the fewer free channels.

When Zhipu GLM-5.2 was just released, Coding Plan users rushed to use it, and the API cost $1.4 per million tokens on a pay-as-you-go basis—not expensive for individual developers, but not a "run freely" price either. After the open-source weights were released, local deployment required a GPU. The window to experience the flagship model for zero cost often lasts only a few days.

But SenseTime's SenseNova Token Plan has turned this window into a door.

During the public beta, GLM-5.2 is completely free to use. It's not the kind of "register and get 1000 tokens" token gesture—it's 1,500 calls every 5 hours, continuously refreshed, and free. And it's not just GLM-5.2—SenseNova 6.7 Flash-Lite, SenseNova U1 Fast, and DeepSeek V4 Flash are all included in the free plan.

01 What SenseTime Token Plan Actually Offers

Let's cut to the chase: Free plan, $0, currently in public beta.

Notice that DeepSeek V4 Flash only has 150 calls—SenseTime resells this channel and pays DeepSeek for each call; 150 times is more like a "taste ticket." But actual output speed reaches 140 token/s, which is quite impressive among free options.

And GLM-5.2's 1,500 calls per 5 hours is a real self-service buffet.

02 Why GLM-5.2 Is Worth Using for Free

On June 13, Zhipu opened GLM-5.2 to all GLM Coding Plan users. Key specs of this model:

• 1M context—currently one of the longest among domestic models, enough to fit the entire "Three-Body Problem" with room to spare
• High/Max dual thinking intensity—recommended Max for programming tasks, deeper reasoning
• MIT open-source license—weights can be used freely, even commercially
• API pricing: Input $1.4/million tokens, output $4.4/million tokens—about 1/7 of Claude's tier

In the SenseTime Token Plan, all these capabilities are completely free. You don't need to buy a Coding Plan or top up API balance; just register and use.

03 How to Get Started: 3 Minutes

Step 1: Register

Visit https://www.sensenova.cn/token-plan , register an account, and select the Free plan.

Step 2: Get an API Key

Enter the console and create an API Key. Supports up to 20 keys.

Step 3: Integrate with Your Tools

SenseTime's API is compatible with OpenAI format. API endpoint:

https://token.sensenova.cn/v1/chat/completions

Claude Code Integration:

export OPENAI_API_KEY="Your SenseTime API Key"
export OPENAI_BASE_URL="https://token.sensenova.cn/v1"

Hermes Agent Integration:

Add a custom provider in config.yaml:

custom_providers:
  - name: sensenova
    base_url: https://token.sensenova.cn/v1
    key_env: SENSENOVA_API_KEY

OpenClaw Integration:

SenseTime officially supports native fast integration with OpenClaw and Hermes Agent, with a one-click configuration entry on the Token Plan page.

Direct Python Call:

from openai import OpenAI

client = OpenAI(
    api_key="Your SenseTime API Key",
    base_url="https://token.sensenova.cn/v1"
)

response = client.chat.completions.create(
    model="glm-5.2",
    messages=[{"role": "user", "content": "Write a quicksort in Python"}],
    stream=True
)

for chunk in response:
    if chunk.choices[0].delta.content:
        print(chunk.choices[0].delta.content, end="")

04 Beyond GLM-5.2: SenseTime's Hidden Cards

SenseTime's strategy is written all over it: Free tokens are the hook; Cowork-Skills are the fish they really want to catch.

SenseNova 6.7 Flash-Lite is not an ordinary chat model—it's a multimodal agent that can directly "read" web page layouts, document structures, financial charts, achieving integrated "see, think, do." SenseTime encapsulates these capabilities into SenseNova-Skills, covering:

• 📊 Data Analysis—automatically generates reports from 900,000 sales records
• 📑 PPT Generation—one-click from outline to layout
• 🖼 Infographic Creation—dedicated to U1 Fast, Infographics generation
• 🔍 Deep Research—8-chapter vertical industry report, autonomous composition

These Skills are open-sourced on GitHub and natively support the Hermes Agent and OpenClaw frameworks.

05 Comparison with Other Free Channels

SenseTime's advantage: truly zero barrier—no subscription needed, just register and use. Disadvantage: the public beta may end at any time, quotas may be adjusted.

06 A Few Notes

1. Public beta does not mean permanent free—SenseTime clearly marks it as "public beta," and Lite/Pro paid tiers will follow later. If you want to free-ride, do it now.
2. GLM-5.2's multiplier—In Zhipu's own Coding Plan, peak hours (14:00-18:00) consume 3x, non-peak 2x. On SenseTime's side, it's currently counted as 1x, effectively saving another layer.
3. Documentation not yet updated—SenseTime's console already shows GLM-5.2 usage, but official documentation hasn't been updated yet; actual usage prevails.
4. LobeChat already supports—If you use LobeChat or similar clients, GLM-5.2 is already visible in the model list.

In Conclusion

A trend in the 2026 AI market is becoming clearer: models themselves are depreciating, while entry points and workflows are appreciating.

SenseTime is giving away GLM-5.2 for free not out of charity—it wants you to get used to the SenseNova ecosystem, accustomed to Cowork-Skills workflows, and used to using SenseTime models within Hermes Agent and OpenClaw. Zhipu giving 20 million tokens to new users follows the same logic.

But for developers, this is precisely the best window. Multiple vendors are competing for entry points, free tokens are flying everywhere, and flagship models are freely available. Once the landscape stabilizes, the free lunch will be gone.

Claim here: https://www.sensenova.cn/token-plan

Promotion: Want a more powerful model? Agnes AI offers 1M context + 4K image generation + video all free, API address apihub.agnes-ai.com/v1 → https://platform.agnes-ai.com/