Fast, general-purpose classifiers like Jev can be used as temporary solutions for specific tasks, then replaced with cheaper, faster custom classifiers trained on the input-output data the Jev instance generated. This creates a practical path for engineering teams without ML expertise to eventually build optimized task-specific models.

• System One models solve the immediate problem of deploying classifiers without ML expertise or large datasets — you just prompt them like an LLM
• Once a System One model proves a feature works, you can collect its decisions as training data to build a specialized classifier that's faster and cheaper to run
• This pattern lets teams validate whether a feature is worth building before investing in custom ML development, reducing risk and upfront cost

The article compares cost-plus and value-based pricing for AI inference resellers, showing how cost-plus margins shrink as inference commoditizes while value-based charges per outcome retain durable margins. It also covers cost-optimization tactics—model routing, caching, distillation—and explains why bring-your-own-key customers break cost-plus but still fit value-based and optimization models.

• Cost-plus pricing on inference collapses as models commoditize since customers can switch to cheaper raw API calls once they spot the markup
• Value-based pricing (Sierra charging per resolved ticket, Devin's Agent Compute Units) decouples revenue from inference costs by charging for outcomes instead
• Distillation—training a sub-8B "student" model from a "teacher" model—can cut per-call costs (e.g., $1.00 to $0.70) while creating a proprietary edge that's harder to copy than caching or routing
• Bring-your-own-key customers break cost-plus pricing entirely but still work under value-based or platform-fee/optimization models

Anthropic quietly throttled its new Claude Fable 5 model with invisible guardrails to block distillation and other high-risk queries. After criticism from researchers and rivals, the company will now reroute those requests to Claude Opus 4.8 and clearly notify users each time a safeguard triggers.

• Anthropic secretly degraded Claude Fable 5's answers when it suspected distillation attempts, without ever notifying users
• After researcher and competitor backlash, Anthropic will now reroute suspected distillation queries to Claude Opus 4.8 with a visible notice instead of silently garbling responses
• Anthropic admits the covert approach was a misstep, chosen originally to ship Fable faster and avoid false positives
• The company still relies on its terms of service banning use of Claude's outputs to train competing models, regardless of whether the throttle triggers