🔍 Read the full analysis: I Asked Jev To Look At 30 Images I Created on ThorstenMeyerAI.com
Get hardware and tech essentials 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
TypeSafe added opt-in image input and bounding-box output to its Jev decision model in a release dated October 14, 2026. Thorsten Meyer tested the features on 30 images from his site; product-image results were strong, but the supplied report cuts off before giving the news-image findings or full test results.
TypeSafe added image input and a bounding-box response type to its Jev decision model on October 14, 2026, and writer Thorsten Meyer tested the new features on 30 images published on his website. In the results provided, Jev 1.14 matched all 10 product images to their own headlines at higher scores than unrelated headlines, but the report ends before giving the remaining results for news and essay images.
Jev is a model designed to answer structured questions rather than generate prose. Before this release, its questions had to concern text. TypeSafe’s update allows an image to be included in the request state, question instructions or criteria for choices and scoring. The image feature works with jev-1.13.0 and jev-1.14.0. A new box response type, which returns one bounding box and a confidence score for a specified target, is limited to Jev 1.14.
Meyer tested the feature against the 30 most recent featured images on ThorstenMeyerAI.com: 10 product roundups, 15 news posts and five essays. For each image, he asked whether it fit its own headline, a randomly selected different headline and an unrelated headline about repotting a plant. The 10 product images also received a multiple-choice question asking which product each image showed. Meyer says he ran about 60 requests across the two Jev versions.
In the reported product-image results, Jev 1.14 gave 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; the plant headline scored no higher than 0.10. Jev 1.14 identified the product correctly in nine of 10 images. The report describes one error involving a mesh Wi-Fi image, but its account of the test stops as it begins the news-image results.
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
Where Image Checks Could Help
The test addresses whether a featured image matches its headline. A structured model that compares images with text could help publishers identify potential mismatches across collections. In this small sample, Jev’s product-image scores were higher for the corresponding headlines than for unrelated ones.
The findings do not establish how reliably the model would perform across other sites or image types. The sample contains 30 images from one publication, and the available account does not supply the outcomes for its 15 news images and five essays. The product test also produced an incorrect answer. The results document one early evaluation and do not establish general performance or production readiness.
As an affiliate, we earn on qualifying purchases.
Opt-In Features With Caveats
TypeSafe says the new capabilities are behind explicit request headers because it considers them experimental. The company warned that there is “still way too much jaggedness” for it to endorse the features for production use. Image requests use the TypeSafe-Danger-Zone-Images: true header; a separate header opts into longer context. TypeSafe said ordinary production behavior remains unchanged unless customers enable the features.
The image request design uses references in an assets map, which are replaced wherever they appear in the request. TypeSafe’s published limits cover PNG and JPEG still images from 256 to 1,568 pixels per side, with a maximum of 16 MiB per image and 32 MiB per request. It says images are billed as ordinary input tokens rather than subject to an additional image charge. The release also raises the optional context limit from 64,000 tokens per request and 32,000 per question to 262,000, while TypeSafe advises caution because reliability may fall with longer inputs.
Jev 1.14 is available as jev-preview. According to TypeSafe, the jev-latest alias is due to move to it about a week after the release, around October 21, 2026. The company characterized the model update as a set of small, broad improvements and said multilingual support is expected to benefit most.
““There is still way too much jaggedness for us to vouch for these for production use cases.””
— TypeSafe, in its release post as quoted by Thorsten Meyer
As an affiliate, we earn on qualifying purchases.
The Missing Test Results
The supplied source text cuts off immediately after the phrase “News images: Jev was right, my images weren’t a dif”. It does not provide the rest of Meyer’s findings, any scores for the 15 news images, results for the five essays, or a complete comparison between Jev 1.13 and 1.14. Those outcomes cannot be inferred from the product-image results.
The account also does not describe an independent evaluation, a larger benchmark or how the model performs on images outside this site’s sample. TypeSafe itself has cautioned that it cannot yet vouch for the new features in production. The specific reliability limits, including how often confidence scores flag mistakes in other datasets, remain unknown from the available material.
As an affiliate, we earn on qualifying purchases.
Jev Rollout and Further Testing
TypeSafe’s stated near-term milestone is the move of jev-latest to Jev 1.14, expected about one week after the October 14 release, or around October 21. The company has also described deprecation notice periods of 60 days for version 1.13.0 and 30 days for version 1.14.0, while saying a formal policy is still to come.
For the image test, the next useful information would be the missing results for news images and essays, along with the complete version-by-version comparison. Until then, the reported product results offer a limited indication of how Jev handled this particular set of images, not a broad measure of its accuracy or suitability for automated publishing decisions.
As an affiliate, we earn on qualifying purchases.
Key Questions
What changed in Jev 1.14?
TypeSafe added opt-in image input and a new bounding-box response type. Image support also works with Jev 1.13.0, while the box type is limited to Jev 1.14.
How did Jev perform on the product images?
In Meyer’s test of 10 product images, Jev 1.14 matched each image with its own headline at scores from 0.65 to 0.89 and identified the product correctly in nine cases. The reported test was conducted on images from one website.
Did Jev perform well on news images?
The available source does not give the news-image findings. It cuts off as Meyer begins that section, so the results for the 15 news images are unknown.
Is TypeSafe recommending the image feature for production?
No. TypeSafe said the new capabilities remain too inconsistent for it to vouch for production use and placed them behind explicit opt-in headers.
When is Jev 1.14 expected to become the latest model?
TypeSafe said jev-latest was expected to move to Jev 1.14 about a week after the October 14, 2026 release, around October 21.
Source: ThorstenMeyerAI.com
Halloween Picks
halloween
As an affiliate, we earn on qualifying purchases.
