TypeSafe AI served one trillion tokens on Modal following Jev launch

MModal10/7/2026, 10:03:22

According to Modal, the TypeSafe AI team served one trillion tokens on its compute infrastructure between Jev's launch on a Thursday and Sunday. Modal also shared a video overview detailing how the team scaled Jev on its platform.

#modal #typesafe-ai #jev #ai-infrastructure

View original post ↗

Previous articleCloudflare AI Gateway standardizes credential rejection responses to HTTP 401 with error code 2009 Next articleEmDash screens plugin listings with Cloudflare's Clef model

Related AI news

DTap red team reports 70.1% direct-misuse ASR on Jev 1.13

Zhaorun Chen says red-teaming Jev 1.13 on DTap found a 70.1% attack success rate for direct misuse and 43.5% for indirect prompt injection. He says using Jev to gate its own tool calls reduced that rate while keeping most of its utility; TypeSafe AI questions whether Jev is failing and says it is built for composability.

TTypeSafe AI•9/28/2026, 20:03:10•

#jev #dtap #prompt-injection #red-team #typesafe-ai

a16z discusses TypeSafe AI’s Jev, a model that returns choices with confidence

Diogo Almeida spoke with a16z’s Ben Horowitz and Martin Casado about Jev, a model built to live inside software. It reads natural language and returns one choice from a set of options, with a confidence level for each.

TTypeSafe AI•9/28/2026, 10:20:00•

#jev #typesafe-ai #a16z #ai-models

New Jev signups no longer get free credits

TypeSafe AI says new Jev signups no longer receive free credits and hopes to restore them as soon as possible. Jev can be started in the TypeSafe console.

TTypeSafe AI•9/27/2026, 15:32:55•

#jev #typesafe-ai #free-credits #signups

Paper uses one Jev yes/no score to flag alignment failures

A paper reports that TypeSafe AI’s Jev, with no extra training, separates alignment failures from good responses at a median AUROC of 0.886. Across 19 benchmarks, one Jev pass cost $0.30, compared with $18.96 for the LLM judges those benchmarks use.

eelvis•9/27/2026, 13:03:40•

#jev #typesafe-ai #alignment #rlcdalignbench #benchmarks

JEV cascade keeps 99% of GPT-6 accuracy at about 57% of the fee

A Jev-as-a-Judge paper reports that accepting TypeSafe AI’s confident JEV verdicts and escalating the rest to GPT-6 Astra kept 99% of GPT-6 accuracy on 510 held-out preference pairs, at about 57% of GPT-6’s fee. The authors say the escalation threshold did not transfer to every fallback model and should be set on one’s own data.

DDAIR.AI•9/24/2026, 08:48:35•

#ai-evaluation #jev #llm-judges #typesafe-ai #gpt-6

Reset history AI News FAQ Usage limits RSS feeds API MCP Skill

News is compiled from public information. The data is for reference only, and this site has no affiliation with OpenAI.

Data & original project: Did Codex Reset · Licoy