Skip to main content
Van Data Team
Current modelMeta

Muse Glimmer benchmarks, pricing and specs

Muse Glimmer is a Meta model released on August 10, 2026 that takes text and image input and produces text. Its price is not published and accepts up to 131,072 tokens of context. It scores 1425 Elo on Arena Text, #41 of 49 current models (Arena, measured October 8, 2026).

Context window
131K
Arena Text
1425 Elo · #41/49

Data verified October 13, 2026 · Dataset 2026-10-13

Share this

Key facts

Provider
Meta
Model family
Muse Glimmer
Release date
August 10, 2026
Status
Current
Access
open weights
API model id
Not reported
Context window
131,072 tokens
Max output
Not reported
Input
text, image
Output
text
Task families
text, multimodal
Declared capabilities
reasoning, coding, agentic coding, image understanding, document understanding

Muse Glimmer pricing

UnitPrice (USD)TierAppliesSource
1M input tokensNot reportedstandardCurrent
1M output tokensNot reportedstandardCurrent
1M input tokensNot reportedcachedCurrent

Independent benchmark results

Scores come from independent evaluators that ran the model themselves, pinned to a dated snapshot. Rank is among current models in the same snapshot. Meta's own launch figures are listed under strengths with the publisher named, because they are not measured the same way.

BenchmarkScoreRankMeasuredSource
Arena Text1425 Elo#41 of 49October 8, 2026Arena · 2026-10-12

Strengths, cautions and best uses

Strengths

  • A 30-billion-parameter open-weight model distilled from Muse Spark, released under Apache 2.0 for local agent workflows
  • Meta reports that 4-bit quantization brings the language model under 20 GB, enough for one consumer GPU or a Mac
  • Meta reports 76.0 on SWE-Bench Verified and 75.5 on MCP Atlas at high reasoning

Cautions

  • Self-hosted only: Meta does not serve it over its Model API, so there is no first-party per-token price
  • Audio is not supported and video is read as individual frames; the default context is 128K tokens

Best for

  • Local or on-device coding agents on a single GPU
  • Fine-tuning an Apache 2.0 multimodal base

Alternatives to Muse Glimmer

The closest current textmodels by price. The cheapest model that scores as well or better is listed first; failing that, the cheapest one within 3% of the benchmark's scale.

Compare Muse Glimmer with

Articles about Muse Glimmer

Muse Glimmer FAQ

What is the context window of Muse Glimmer?

Muse Glimmer accepts up to 131,072 tokens of context. That ranks #51 of 57 current text models by context length.

How does Muse Glimmer score on independent benchmarks?

Muse Glimmer scores 1425 Elo on Arena Text, #41 of 49 current models (Arena, measured October 8, 2026). These are independent evaluations, not Meta's own launch figures.

Is Muse Glimmer open source or API only?

Muse Glimmer is released as open weights that you can run yourself.

What are the best alternatives to Muse Glimmer?

The closest current alternatives by price and measured performance are Nemotron 3.5 Lightning (76 points lower on Arena Text, 1349 vs 1425), Grok 4.6 (29 points higher on Arena Text, 1454 vs 1425), DeepSeek-V4-Pro (40 points higher on Arena Text, 1465 vs 1425) and Gemini 3.7 Flash (61 points higher on Arena Text, 1486 vs 1425).

Sources

Cite this page

Van Data Team. "Muse Glimmer Benchmarks, Pricing and Specs." AI Model Benchmark Dataset 2026-10-13. https://vandatateam.com/ai-model-benchmarks/muse-glimmer

Reuse the data: JSON · CSV · or explore every model in the AI model benchmark comparison.

Production evaluation

Deciding whether Muse Glimmer fits your workload?

Public benchmarks narrow the shortlist. We test the finalists on your own tasks, data and cost limits, then build the routing and review controls to run the winner in production.

  • Evaluation cases and acceptance thresholds tied to real work
  • Model routing, fallbacks and cost controls
Compare all modelsDiscuss your evaluation