Pega Infinity 26 eliminates AI token tax with predictable costs
Pegasystems Inc. announced Pega Infinity 26 at PegaWorld, introducing a pricing model that charges per completed case instead of per token. The Pega Predictable AI architecture shifts reasoning to design time to ensure predictable outcomes and lower costs. New pre-built agents and an AI Token Cost Calculator were also introduced.

*this image is generated using AI for illustrative purposes only.
Pegasystems Inc. announced at PegaWorld that clients can now design, build, and run agentic workflows across Pega Infinity 26 without paying per token. The Pega Predictable AI architecture shifts heavy AI reasoning to design time, ensuring runtime agents are fast, reliable, and cheaper to run. This approach addresses escalating token costs and unreliable outcomes, offering a flat price per completed case regardless of AI usage behind the scenes.
Market Context and Challenges
Token bills are shocking enterprise leaders as LLM providers convert flat-rate subscriptions to expensive token-metered pricing. Complex requests require more reasoning steps, increasing costs and the likelihood of inconsistent answers. Gartner reports that over 40% of agentic AI projects may be canceled by 2027 due to rising costs and unclear business value. Pega's outcomes-based model charges per completed task, aligning costs directly with business value.
Architectural Difference
Pega applies AI reasoning at design time using Pega Blueprint AI and Pega Infinity Studio to reimagine processes. Once deployed, the system shifts to a lightweight semantic mode for runtime. Agents use a lightweight AI query to understand user intent and follow pre-approved workflows step-by-step. This eliminates the need to re-reason each new workflow, ensuring consistency and efficiency.
Key Benefits
The new architecture delivers two critical advantages:
- Predictable outcomes: Agents follow pre-approved workflows consistently, which is critical for regulated industries.
- Predictable costs: AI reasoning occurs once at design time, not repeatedly at runtime, making it more efficient and affordable.
To quantify savings, Pega introduced an AI Token Cost Calculator. This interactive tool estimates potential savings by comparing Pega AI with token-metered alternatives, with some clients realizing savings of more than 20x.
New Pre-Built Agents
Pega introduced pre-built agents to automate workflows further:
| Agent Type | Functionality |
|---|---|
| Agentic Assignment Agent | Engages employees or customers via email, chat, or telephony for data or approvals. |
| Document Agent | Analyzes complex documents by splitting, categorizing, bounding, scoring, and triaging them. |
Availability and Industry Support
Pega Infinity 26 will be available in Q3. Industry leaders emphasized the importance of these developments. Alan Trefler, founder and CEO of Pega, stated that charging by meaningful work accomplished rather than token usage provides organizations freedom to use AI agents. Liz Miller of Constellation Research noted that solutions consuming tokens with high efficiency will provide a competitive advantage.
How will competitors in the enterprise software space respond to Pega's outcomes-based pricing model?
Will the shift to design-time reasoning limit the flexibility of agents to handle novel, unforeseen scenarios at runtime?
What impact will this pricing strategy have on the margins of major LLM providers if other vendors adopt similar token-avoidance architectures?




























