Why vision input Changes the Economics of Discovery

For most of the last year, the conversation around fine-tuned distillation has been louder than the evidence. That is starting to change.

Duolingo has been quietly running onboarding through Claude Projects for months. The results are unglamorous and, for that reason, more interesting than another benchmark chart.

Eval harnesses, once an afterthought, are becoming the most important piece of code in many AI projects. Zendesk's team treats theirs the way an SRE team treats a runbook.

The cost curve matters here. Llama 4 is roughly an order of magnitude cheaper per token than the equivalent model 18 months ago, and that changes which problems are worth automating at all.

The skeptical read is that we are watching a feature, not a platform. The optimistic read is that vision input is exactly the kind of feature that becomes a platform when nobody is paying attention.

What Anthropic actually shipped with Claude 4.5 Sonnet is less a single capability and more a cluster of small, compounding improvements — the kind that only show up when you put a real workflow on top.

Eval harnesses, once an afterthought, are becoming the most important piece of code in many AI projects. Klarna's team treats theirs the way an SRE team treats a runbook.

Inside Atlassian, the rollout looked less like a moonshot and more like a slow migration. A pilot, a champion, a quiet expansion, a budget line.

None of this guarantees a clean story. Inflection could ship a model next month that rearranges the assumptions in this piece. But the direction of travel, for now, is clear enough to plan around.