Why high: one template-boundary shift from a group whose members all declare the same family.
Meta: Muse Glimmer 30B is Meta's model. Its measured tokenizer sits with the Llama4 group, which is evidence that it builds on the Llama4 tokenizer. Tokenizer reuse is normal engineering, open vocabularies travel between labs, and this says nothing against Meta's authorship of the model itself.
| We checked | What we found | Where it came from |
|---|---|---|
| How it counts tokens | Counted cleanly, and no other model in the catalogue counts the same way | tk_20b9d005 |
| How its API is set up | A combination of settings no other listing in the catalogue uses | api_0a22d378 |
| How much it can read and write | Reads up to 131,072 tokens at onceOpenRouter's listing shows 117,964 output tokens, which is 90% of the context window, the figure it carries where the provider declares no maximum. | its own listing |
| Whether it thinks before answering | Always reasons, and cannot be turned off, at xhigh, high, medium, low, medium by default | its own listing |
| Who served our requests | Fireworks | we measured it, 21 August 2026 |
Click a row to open the full comparison.
We sent Meta: Muse Glimmer 30B 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: 56 prompt tokens.
Muse Glimmer 30B is a dense, open-weight multimodal model from Meta Superintelligence Labs, distilled from Muse Spark and optimized for autonomous agents on consumer hardware. It is suited for long-horizon...
Supported parameters
frequency_penaltyinclude_reasoninglogit_biasmax_tokensmin_ppresence_penaltyreasoningreasoning_effortrepetition_penaltyresponse_formatseedstopstructured_outputstemperaturetool_choicetoolstop_ktop_p
defaults: {"temperature":1,"top_p":0.95,"top_k":64}
