New playbook: when your own AI model pays. Read
DoorDash (delivery marketplace, ~20,000+ employees) reported this on 6 Jul 2026 in company self-report (source). It belongs to the models and routing layer.
A staged code-review agent (cheap scout model flags areas, stronger reviewer verifies) plus DashBench, which replays historical PRs to compare model mixes on recall, precision, cost and latency.
Production reviewer (Sonnet 4.6 scout + Opus 4.8 reviewer): 53.6% weighted recall vs 30.7% for a single-pass GPT-5.5 baseline, at $3.91/PR. Swapping the scout to open-weight Kimi K2.6: $2.35/PR at 45.8% recall.
The number covers 105-case DashBench report on historical PRs.
“Best" is meaningless until you say best at what, on which cases, at what cost,” DoorDash said.
One gateway picks the AI model for each task and records what it costs.
Up to 90%
lower AI cost, AT&T says
An internal AI Gateway that, turn by turn, weighs speed, cost and expected quality and routes each prompt to the most cost-effective model, switching models mid-session, plus a library of post-trained open telco models (OTel 2.0) that run on-prem or in cloud (source).
<8%
of closed-model cost per transaction
A blended model portfolio (in-house compact models, open-source models post-trained on Pinterest data in its own secure cloud, closed models only in 'very limited cases') behind a centralized routing layer that sends routine work to lighter models; being extended to internal engineering and sales tools in H2 2026 (source).
52%
lower cost per agent session since the June peak
A cost equation and weekly metric stack for its agentic 'software factory' (70%+ of PRs agent-attributed), with benchmark-driven Pareto model selection (incl. open-weight models), cheaper default subagent models, Medium reasoning effort default, 400K-token compaction cap and prompt-cache strategy (source).