TypeSafe AI launches Jev on Merge Gateway

// 1h agoPRODUCT UPDATE

TypeSafe AI launches Jev on Merge Gateway

Merge Gateway has integrated Jev, a specialized decision engine developed by TypeSafe AI that is purpose-built for machine-to-machine software workflows rather than conversational text generation. By processing unstructured application state alongside typed queries, Jev yields structured decisions with calibrated probabilities at minimal latency, eliminating the computational and financial overhead of traditional autoregressive LLMs. Available on Merge Gateway at $0.042 per million input tokens with free output tokens, the model provides an economical and reliable option for automated classification, routing, and guardrail pipelines.

// ANALYSIS

Replacing general-purpose autoregressive LLMs with dedicated, typed decision models is the natural evolution for high-throughput production infrastructure.

• Eliminating verbose token generation dramatically cuts latency and compute costs for tasks that only require a structured judgment.

• The $0.042/M input pricing paired with free output tokens makes large-scale automated routing and data labeling economically practical.

• Distribution via Merge Gateway provides engineering teams with immediate, standardized access without the friction of self-hosting or bespoke API integrations.

// TAGS

typesafe-aijevmerge-gatewayai-infrastructuredecision-modelsapi-gateway

DISCOVERED

1h ago

2026-09-18

PUBLISHED

1h ago

2026-09-18

RELEVANCE

7/ 10

AUTHOR

merge_api

// KEEP READING

More AI developer news from the feed

EXPLORE FULL FEED

MODEL25m ago

Jev Breaks AI Gateway Adoption Record

TypeSafe AI’s Jev reached roughly 13% of teams on Vercel AI Gateway in its first day—about twice the GPT-5.6 family and six times Fable 5.1. Jev returns typed, probabilistic decisions instead of generated text.

RESEARCH35m ago

NVIDIA's SoL-Pi Nearly Halves Agent Token Traffic

NVIDIA's SoL-Pi is an open-source Pi coding-agent extension that fuses actions, compresses observations, preserves diagnostic evidence, and compacts context. On 51 EdgeBench tasks, it matched Pi closely across GPT-5.6 Sol and Opus 5 while cutting token traffic 44.7–49.0% and API costs by roughly one-third.

LAUNCH44m ago

Claude for Small Business tackles back-office toil

Anthropic’s Claude for Small Business packages ready-to-run workflows and app connectors inside Claude Cowork, helping owners handle finance, operations, sales, marketing, HR, and customer service tasks. Approval gates and inherited permissions keep users in control before Claude sends, posts, or pays.