New token model for Treble SDK

From September 1st, 2025, the Treble SDK moves to an itemized token model that replaces the time-based approach. The cost of each job is now known in advance and shown in the SDK based on the job inputs rather than an estimate of simulation time. This makes pricing transparent, aligns incentives with faster solvers, and enables value-based pricing. Use the token estimator to explore how parameters scale, while the SDK always displays the final cost computed from the mesh.

The new model introduces an itemized token structure, replacing the previous time-based model. Under this updated system, the token cost of a compute job will be deterministically known in advance, based on the job’s input parameters, removing the reliance on estimated simulation times.

Motivation Behind the Model Update

Treble is committed to transparency and alignment of incentives. The transition to an itemized model is driven by three key motivations:

• Greater transparency and predictability: Customers can now calculate the exact token cost of a job before execution, eliminating uncertainty tied to simulation time estimation.
• Alignment of incentives: Under the previous model, improvements in solver speed inadvertently reduced revenue. The itemized approach ensures Treble is fully incentivized to make its simulation technology as fast and efficient as possible.
• Value-based pricing: Treble aims to align pricing with the actual value delivered to customers. By decoupling price from simulation duration, the new model supports more flexible, feature-driven pricing and empowers us to prioritize high-impact capabilities that deliver the most value to our clients, not merely those that consume the most compute.

Key Practical Changes

• Wave-based simulations: The pricing is de facto unchanged from the previous model.
• GA simulations: The pricing has been revised (with some cases higher and others lower), but overall token costs for GA jobs remain very low.
• Receiver placement: A small cost will be associated with placing a receiver. For mono receivers the cost is virtually 0, whereas a higher ambisonics receivers comes with a small fee.
• DRTF renderings: Now incur a small token fee and are executed server-side, enabling efficient batch processing.
• Deterministic pricing: Token costs are fully known in advance within the SDK for all simulation and rendering jobs.

You can use our new token estimator to get a general sense of how simulation cost scales with different parameters. Please note that the estimator provides an approximate value, the actual token cost is determined by the mesh and is calculated within the SDK. The cost displayed in the SDK is the final, accurate cost of the simulation.