AI Desk · AI FingerprintingNew arrival
Why high: an exact fingerprint shared with 22 models that declare the same family, without a second signal of a different kind yet.
Elorian: Glimpse is Elorian's model. Its measured tokenizer sits with the Qwen group, which is evidence that it builds on the Qwen tokenizer. Tokenizer reuse is normal engineering, open vocabularies travel between labs, and this says nothing against Elorian's authorship of the model itself.
Elorian says Glimpse was "post-trained on an open weights base model". Glimpse reads with Qwen3.8's tokenizer and wraps every prompt in Qwen3.8's chat template. Qwen3.8 27B's repository tokenizer and template, run over the same 50 strings offline, return the served counts on all 50, and the template's thinking instruction at its default effort, a system message opening "Reasoning effort is set to xhigh", accounts for the 52-token baseline exactly; with thinking off, its prompts are 40 tokens shorter, as the template's are. Five more prompts, each sent with thinking on and off, matched the template's length to the token in all ten requests. Asked who made it, in English and in Chinese, every answer names Qwen or Alibaba Group's Tongyi Lab. Qwen3.8 27B is an open-weight vision model Alibaba released under the Apache 2.0 licence.
Elorian, Introducing Glimpse (16 October 2026)Elorian team pageQwen3.8-27B chat template (the reasoning-effort system message)Qwen3.8-27B repository (tokenizer.json; config.json: vocab_size 248320; Apache 2.0)Glimpse on OpenRouter
| We checked | What we found | Where it came from |
|---|---|---|
| How it counts tokens | Counts every one of the 50 test strings exactly like 22 other models | tk_181907f2 |
| How its API is set up | A combination of settings no other listing in the catalogue uses | api_fc0a703c |
| How much it can read and write | Reads up to 262,144 tokens at once, writes up to 256,000 | its own listing |
| Whether it thinks before answering | Reasons when asked to | its own listing |
| Who served our requests | Together | we measured it, 17 October 2026 |
Click a row to open the full comparison.
We sent Elorian: Glimpse fifty short pieces of text and recorded what each one cost it in tokens. The bars below are those costs. Two models built on the same tokenizer produce the same bars; a model built on a different one produces a different set, which is what makes this a fingerprint.
Overhead subtracted: 52 prompt tokens.
Glimpse is a multimodal visual reasoning model from Elorian, built for visual understanding and complex visual reasoning in the physical world. It reasons directly over images and video rather than...
Supported parameters
frequency_penaltyinclude_reasoninglogit_biaslogprobsmax_tokensmin_ppresence_penaltyreasoningrepetition_penaltyresponse_formatstopstructured_outputstemperaturetool_choicetoolstop_ktop_logprobstop_p
