Is this Jev site official? TypeSafe domains vs lookalike resellers, with prices
Within a week of Jev's launch, hundreds of new "jev" domains appeared. Some resell real Jev access at several times TypeSafe's price, some are information sites or playgrounds, and none of them is TypeSafe. This page lists the official surfaces, a five-step check, and what each reseller displayed on 28 September 2026. One more lookalike candidate, thejevai.com, was checked on 7 October 2026.
TypeSafe's own surfaces are typesafe.ai and its subdomains, the github.com/typesafe-ai organisation, and the packages typesafe-sdk (PyPI) and @typesafe-ai/sdk (npm). The official price is $0.042 per million input tokens with free output. Eye Security (25 Sep) found lookalike reseller storefronts charging 3–11.5× that price; on 28 Sep one pack on jevmodel.org displayed 15.7×. Eye Security wrote that this "isn't illegal as far as we can tell" and did not call the sites malicious.
Highest displayed price as a multiple of TypeSafe's $0.042 per million input tokens, 28 Sep 2026
Five-step check
- Domain. Is it
typesafe.aior a subdomain (docs.,console.,api.,status.)? Eye Security's advice: "check that you're on console.typesafe.ai, not just a site with Jev in its name". - Packages and code. Is it
typesafe-sdkor@typesafe-ai/sdk, with code undergithub.com/typesafe-ai? Thegithub.com/TypeSafeAIorganisation, the PyPI packagetypesafe-aiand the unscoped npm packagen8n-nodes-typesafe-aiare not TypeSafe's; TypeSafe's n8n node is the scoped@typesafe-ai/n8n-nodes-typesafe-ai. - Key. Does the key come from
console.typesafe.aior a named gateway (OpenRouter, Vercel, Cloudflare, Netlify, Opper)? A reseller's key is not a TypeSafe key; jev-agent.org, for example, "provisions a Jev Agent key rather than an official TypeSafe API key". - Price. The official price is $0.042 per million input tokens and output is free. Any higher per-token price is a markup.
- Operator and data. Does the site name its operator and state any affiliation? What does it ask for: card, key, prompts? Eye Security found reseller traffic passing "an app on Railway behind Cloudflare" with "no contract or SLA". Never paste your TypeSafe key into a third-party site; jev.works asks for one. How official routes handle your prompts, by contrast, is quoted on data handling by route.
Official TypeSafe surfaces
| Surface | Official value | How the pool confirmed it |
|---|---|---|
| Website and docs | typesafe.ai, docs.typesafe.ai | Homepage links the docs |
| Console, keys, playground | console.typesafe.ai | Homepage "Sign in"; docs quickstart |
| API | api.typesafe.ai (/v1/systemone, /v1/models) | API reference and models page |
| Status page | status.typesafe.ai | On TypeSafe's domain; not linked from the website or docs on 28 Sep |
| GitHub | github.com/typesafe-ai (SDKs, skills, System One adapter) | Linked from the docs quickstart and the PyPI and npm repository fields |
| Python package | typesafe-sdk 0.7.2 (26 Sep 2026) | Docs SDK page; PyPI author email on the typesafe.ai domain |
| JavaScript package | @typesafe-ai/sdk 0.6.0 (15 Sep 2026) | Docs SDK page; maintainers use typesafe.ai addresses |
| Agent skill | typesafe-ai/skills | Docs quickstart |
| n8n node (added 30 Sep; npm checked 3 Oct) | github.com/typesafe-ai/n8n-nodes-typesafe-ai: MIT, created 23 Sep 2026, tag 0.9.0, no GitHub releases, last push 29 Sep 2026, 21:11 UTC. On npm: @typesafe-ai/n8n-nodes-typesafe-ai 0.9.0 | Repository in the official typesafe-ai organisation (checked 30 Sep 2026, 20:48 UTC). TypeSafe publishes the node on npm as @typesafe-ai/n8n-nodes-typesafe-ai 0.9.0 (provenance to typesafe-ai tag 0.9.0); the unscoped n8n-nodes-typesafe-ai is a third party's package (maintainer dommonte). Documented npm records checked 3 Oct 2026, 05:20–05:27 UTC. Which node to install and what each does when Jev fails: Jev n8n nodes compared. |
| Benchmark code (added 30 Sep) | github.com/typesafe-ai/WorkflowEvals: Apache-2.0, created 28 Sep 2026, one commit on 29 Sep 2026, 16:25 UTC; code behind evals.typesafe.ai | Repository in the official organisation; README read at commit 0ac3b8a (30 Sep 2026, 20:49 UTC). What it measures: see benchmarks, row B01. |
| Social and email | X @typesafeai; CEO @CompleteSkeptic; sales and support addresses on the typesafe.ai domain | Homepage, docs and team page |
| Not official | github.com/TypeSafeAI: "This is an UNOFFICIAL organization made for community repos" (created 18 Sep 2026). PyPI typesafe-ai: a third-party redirect package. | GitHub organisation description; PyPI metadata |
Documented Rows checked 28 Sep 2026, except the two rows added on 30 Sep 2026 (checked 20:47–20:49 UTC); the n8n row's npm publisher was checked on 3 Oct 2026, 05:20–05:27 UTC. The official organisation had 11 repositories on 30 Sep. The billing entity for direct accounts was not checked.
Reseller prices as displayed
Multiples are the pool's own arithmetic: displayed price ÷ $0.042 per million. Prices are what each site displayed on 28 Sep 2026, 19:46–19:47 UTC. The thejevai.com row was added on 7 Oct 2026 from its homepage only, so it shows no price. Reported (sites' own displays)
| Domain | Displayed price per million tokens | Multiple of official | Affiliation statement | Asks for | Named by |
|---|---|---|---|---|---|
| jevtypesafeai.com | $0.42 ($5–$20 packs), $0.35 ($50), $0.25 ($100); output free. Unchanged since 24 Sep. | 10.0×, 8.3×, 6.0× | "Not affiliated with or endorsed by TypeSafe AI"; operator CODEFASHION TECH LTD | Account, card, prompts via its own API | Eye Security, 25 Sep; GBHackers, Cyber Security News, Mallory, 28 Sep |
| jev-ai.pro | Annual plans $0.242, $0.148, $0.124 (monthly prices not shown in the default view) | 5.8×, 3.5×, 3.0× | "an independent playground and API" | Account, card, site key, prompts | Eye Security (one of six sites sharing code) |
| jev-agent.org | Monthly $0.483, $0.297, $0.247; annual $0.242–$0.124; packs $0.427–$0.283 | up to 11.5× | "not affiliated … None of these are TypeSafe's own prices" | Plan purchase; issues a "Jev Agent key" | Eye Security; GBHackers |
| jevmodel.org | Packs $0.66, $0.52, $0.38, $0.28; annual $0.24–$0.12 | up to 15.7× | "Not affiliated with TypeSafe" | Pack or plan purchase; prompts | Eye Security (domain list) |
| jevapi.pro | Per-decision credits, e.g. $990 a year for 120,000 | Not computable (priced per decision) | "through OpenRouter. Not affiliated with TypeSafe." | Plan purchase; prompts | Eye Security |
| jevai.site | No Jev access sale seen on the homepage; shows the official price | — | "an unofficial reference site" | — (sponsored links) | Eye Security lists it; the pool's check did not confirm a sale |
| thejevai.com Lookalike candidate, added 7 Oct 2026. HTTP 200, title "Jev AI — Jev Model Online, Free, No Waitlist" (read ) | No price observed: only the homepage was read. It says "Jev playground runs are free. Laya runs consume account credits." and has a "Pricing" link (not opened) | Not computable (no price read) | Footer: "Jev AI is an independently operated product and is not affiliated with, operated by, or endorsed by TypeSafe." Reported (the site's statement about itself) Observed by the pool: not linked from typesafe.ai's homepage or the docs pages searched; served from Cloudflare addresses that differ from typesafe.ai's (different hosting, not proof by itself). Documented | Not checked (homepage only); a playground with a "Get Started" link. Lists other decision models (Laya, Clef, pplx-decider and others) beside typesafe/jev-1.13 | GitHub repository abaljeu/jev-mcp ("Patched jev-mcp for thejevai.com"). Not named in the Eye Security report |
thejevai.com source: R10-S42, its homepage (05:19:18 UTC) and DNS and response headers (05:24:14 UTC) on 7 Oct 2026; TypeSafe's homepage, docs llms.txt, legal.md and quickstart searched for the domain at 05:19 UTC. The pool did not sign up, pay or submit anything. Details: 7 Oct research report.
jev-ai.pro also offers "Jev-Omni" for images, audio and video. TypeSafe's Jev 1.13 accepts text only, so that product is not TypeSafe's model. Eye Security also named jev-agent.com, which the pool did not check.
What the security researchers actually said
The primary source is Eye Security, "Rise of the Jev-Clones" (25 Sep 2026). GBHackers, Cyber Security News and Mallory (28 Sep) report on it. Reported
- About 670 new "jev" domains received certificates between 15 and 22 Sep, and about 40 "typesafe" domains between 16 and 21 Sep.
- The storefronts sell real TypeSafe Jev with a markup of 3× to 11.5×.
- Eye calls them "fake storefronts" and a "rip-off", but also writes: "This isn't illegal as far as we can tell."
- Cyber Security News: the report "does not describe a malware infection or establish that operators stole prompts."
This site therefore calls them lookalike reseller storefronts (Eye Security, 25 Sep). No named security source calls any of these domains malicious. jevtypesafeai.com also has a "Jev on AWS Bedrock" page; Jev is not listed in the Amazon Bedrock model cards at 5 Oct 2026, 05:18 UTC (Is Jev on Amazon Bedrock or Azure AI Foundry?).
Recheck, 30 Sep 2026, 20:53 UTC: the Eye Security report is still online and still says "You just pay up to 11.5x the official price"; jev-agent.org, jev-ai.pro and jevtypesafeai.com all answered HTTP 200. Their prices were not re-read, so the prices above keep their 28 Sep date. Reported
Other Jev sites that do not resell tokens
jev-ai.live, jevaiplayground.com, jevplayground.com, typesafe.pro, typesafeai.app and jev.works are information sites, playgrounds or gateways; each states it is independent of TypeSafe. jev-router.com is an independent gateway for open "Jev-class" models and states: "We do not provide access to TypeSafe's Jev model." It is unrelated to OpenRouter's typesafe/jev-router. jev.bevel.software ("EU/Jev") is a Jev-compatible API that serves another model: it states that "eu/jev runs its own model" (jeff-latest), that Jev model IDs return HTTP 400, and that it is "Not affiliated with TypeSafe AI"; it claims inference in Germany and a price of €37 per billion input tokens. It does not resell Jev and does not claim to host Jev, so it is a compatible substitute, not Jev. Reported the operator's own pages, read 3 Oct 2026, 05:30 UTC Also flagged but not opened: the GitHub repository jev-ai-desktop/Jev-AI-Desktop (its description is a keyword list), Hugging Face Spaces under jevaimodel/ and a model named JEV-27B-GGUF. TypeSafe's Jev weights are not public, so any download claiming to be them should be treated with caution. Unverified
What was not verified
- Who operates each domain: the pool relies on the sites' own statements and Eye Security's report.
- jev-agent.com, jev-ai.pro's monthly prices, and the full text of the cyberpress article.
- The flagged repositories and Spaces were not opened. No purchases were made and no prompts were submitted.
- Displayed prices can change; each carries the 28 Sep check time.
- thejevai.com: only the homepage was read. Its pricing page, sign-up flow and operator were not checked; the domain registrant was not looked up.