🔍 Read the full analysis: Jev Can See Now: A Hands-On Test With My Own Pictures on ThorstenMeyerAI.com
Get the little things that make your day delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
TL;DR
Thorsten Meyer tested TypeSafe’s Jev image features on 30 images from his website after the company released Jev 1.14 on October 14, 2026. In his reported results, the model separated product images from mismatched headlines and identified 9 of 10 products correctly, but the supplied source cuts off before giving results for news and essay images.
TypeSafe’s Jev decision model can now take images as input, and a hands-on test published by Thorsten Meyer on October 15 found that Jev 1.14 matched product photos to their headlines in the author’s sample. Meyer said it identified the product in 9 of 10 images; the company has cautioned that the new capabilities remain too unreliable for production use without further testing.
TypeSafe released Jev 1.14 on October 14, adding image input to its typed decision model and a new answer type that returns a bounding box around a described target. Image input also works with Jev 1.13, according to the company. Both features require explicit opt-in request headers, which TypeSafe groups under a “danger zone” label.
Meyer tested the image feature against 30 featured images from his site: 10 product roundups, 15 news posts and five essays. He asked whether each image fit its own headline, a randomly selected other headline, and an unrelated headline about repotting a houseplant. Product images also received a multiple-choice question asking which of 10 products each image showed. Meyer said he ran about 60 requests across Jev 1.14 and 1.13.
For the 10 product images, Meyer reported that Jev 1.14 assigned each image’s own headline a score from 0.65 to 0.89, with a median of 0.76. Random other headlines scored no higher than 0.18, with a median of 0.06, while the plant headline stayed at or below 0.10. Jev 1.14 named the right product in nine cases. The provided source ends as it begins describing results for news images, so it does not supply the remaining findings from the test.
Jev 1.14 · images + box
Jev can see now. Here is what 30 of my images showed.
I asked about every featured image three times: does it fit its own headline, a random other post’s headline, an off-topic headline? Both Jev versions, about 60 requests.
Does this image fit the headline?
Median “yes” probability on Jev 1.14, by kind of post
- Product roundups10 images0.760.061.13: 0.79 / 0.05
- News posts15 images0.330.161.13: 0.51 / 0.25
- My essays5 images0.510.161.13: 0.62 / 0.33
Off-topic headline (“How to Repot a Monstera Plant”): never above 0.10 on either version, 60 of 60 rejected.
Why news is weak: almost all news images show the same scene, people at desks looking at monitors. Jev could not tell them apart, and it was right not to.
1.14 is stricter than 1.13
Same 30 images, same questions
| Result | 1.14 | 1.13 |
|---|---|---|
| Fits own headline (≥ 0.5) | 15/30 | 23/30 |
| Fits a wrong headline (≥ 0.5) | 1/30 | 4/30 |
| Own beats random other (AUC) | 0.91 | 0.89 |
| Names the product (10 options) | 9/10 | 10/10 |
1.14’s one product miss: mesh system read as “Wi-Fi 7 router”, confidence 0.60. Re-tune thresholds before jev-latest moves.
What shipped on Oct 14
- headerImagesin state, instructions or criteria · 1.13 + 1.14 ·
TypeSafe-Danger-Zone-Images - 1.14 onlyBox typeone bounding box + confidence. 6 of 6 clear targets boxed (0.66–0.90); an absent bicycle still got a box (0.03)
- headerLong context64k / 32k → 262k tokens, slower and less accurate
- opt-inHelp find weaknessesqueries used for testing, not training
now: jev-preview = 1.14~Oct 21: jev-latest → 1.14notice: 60 days (1.13), 30 days (1.14)
Source: own test on October 15, 2026. 30 featured images from ThorstenMeyerAI.com (10 product roundups, 15 news posts, 5 essays), JPEG at up to 1,568 px, jev-1.14.0 and jev-1.13.0 with the image header. Bars show medians. AUC = probability that an image’s own headline scores higher than a random other headline. Features, prices and limits: TypeSafe blog post “Jev 1.14” and docs.typesafe.ai, as of October 15, 2026.
Powered by Thorsten Meyer AI
Testing Image Checks on Real Site Photos
The release extends Jev from decisions about text to questions that include visual material. That could support narrow, repeatable tasks such as checking whether a featured image fits a headline or locating a named object in a picture. Meyer’s results offer an early example of how such a workflow might be evaluated using a publisher’s own images rather than a public benchmark.
The test is small and comes from one site, so it does not establish how Jev performs across different image types, publishers or operational settings. Its product-image results are encouraging within that sample, but the distinction between a high score and a dependable automated decision remains important. TypeSafe itself says the image and bounding-box features are not ready for production use without explicit opt-in.
As an affiliate, we earn on qualifying purchases.
Jev’s New Image Input Options
Jev is designed to answer structured questions rather than compose prose. Users provide a state and typed questions; the model can return a yes-or-no probability, choose among options or score a response, along with confidence information. Before this release, Meyer said, those questions had to concern text.
TypeSafe’s request format uses an asset map: an image is assigned a reference, and that reference can be placed in the state, question instructions or criteria. Meyer’s account says placement affects how often the image is sent for processing. The company supports PNG and JPEG still images, subject to file and dimension limits; unsupported formats such as WebP and HEIC must be converted. Images are billed as ordinary input tokens, according to the source.
Jev 1.14 also raises the maximum context size through a separate opt-in header, while TypeSafe retains the previous default and advises caution because reliability can decline with longer inputs. The release is initially available as jev-preview; the source says jev-latest was expected to move to it about a week after release.
“The company said Jev 1.13.0 “was multimodal all along.””
— TypeSafe, as described in Meyer’s report
How Far the Test Results Extend
The supplied source material cuts off during the section on news images. It does not include the reported outcomes for the 15 news posts or five essays, the full comparison between model versions, or the conclusion of Meyer’s test. Those findings cannot be inferred from the product-image results.
The test covers only 30 images from one website, and the source does not provide a complete evaluation method, independent replication or evidence about performance in production. TypeSafe also says its internal comparisons are limited and may favor Jev. The reliability of bounding-box answers and the effect of image placement across larger workloads are not established by the material provided.
Preview Rollout and Further Evaluation
TypeSafe made Jev 1.14 available through jev-preview and said the jev-latest alias was due to move to the release around October 21, 2026. The source says the company planned to provide 60 days’ deprecation notice for Jev 1.13 and 30 days for Jev 1.14, with a formal policy still to come.
For users considering image-based checks, TypeSafe’s stated position is to opt in cautiously and test the model against their own material. Meyer also recommends running independent evaluations rather than relying on the company’s internal intelligence estimates. The source does not describe a later production-readiness date or provide additional results from the unfinished test account.
Key Questions
What changed in Jev 1.14?
TypeSafe added image input and a new bounding-box answer type to Jev 1.14. Image input also works with Jev 1.13, according to the company.
What did Meyer’s image test find?
For 10 product images, Meyer reported that Jev 1.14 matched each photo to its own headline with scores from 0.65 to 0.89 and correctly identified nine products. The supplied source does not include the rest of his results.
Is TypeSafe recommending the new image features for production?
No. TypeSafe said the features have too much “jaggedness” for it to vouch for production use and requires users to opt in with explicit headers.
Which image formats does Jev support?
The source says Jev accepts PNG and JPEG still images, within specified size and dimension limits. WebP and HEIC files need to be converted first.
Source: ThorstenMeyerAI.com
Halloween Picks
halloween
As an affiliate, we earn on qualifying purchases.
