ModelBench
Meta logo
Models/Meta

Meta: Muse Glimmer 30B

Muse Glimmer is a 30-billion-parameter open-weight multimodal model from Meta Superintelligence Labs, distilled from Muse Spark for always-on local agents, tool use, coding, and image understanding.

Source-linked muse Open weights Apache 2.0 Released 2026-08-10
API record Report
InputT
OutputT
Input price$0.2/M
Output price$0.8/M
Context131.072K
Max output131.072K
Providers12
Inference availability

Providers

Provider-specific identifiers, limits, and listed prices per million tokens. Every row links back to the provider's own documentation.

Report a price
Providers offering Muse Glimmer 30B
ProviderProvider model IDContextMax outputInputOutputCache readCapabilitiesDocs
Requesty muse-glimmer-30b 131.072K 20.48K — — — ReasoningTools Docs ↗
Nvidia meta/muse-glimmer-30b 131.072K 131.072K — — — ReasoningToolsJSON Docs ↗
EmpirioLabs AI muse-glimmer-30b 131.072K 32.768K $0.2 $0.8 $0.05 ReasoningToolsJSON Docs ↗
Eden AI deepinfra/meta-models/Muse-Glimmer-30B 131.072K 131.072K $0.3 $1.2 $0.04 ReasoningToolsJSON Docs ↗
Kilo Gateway meta/muse-glimmer-30b 131.072K 117.964K $0.3 $1.1 $0.04 ReasoningToolsJSON Docs ↗
OpenRouter meta/muse-glimmer-30b 131.072K 117.964K $0.3 $1.1 $0.04 ReasoningToolsJSON Docs ↗
Eden AI flexai/Muse-Glimmer-30B 131.072K 131.072K $0.3 $1.1 — Reasoning Docs ↗
Eden AI together_ai/meta-models/Muse-Glimmer-30B 131.072K 131.072K $0.35 $1.5 $0.04 ReasoningToolsJSON Docs ↗
Vercel AI Gateway meta/muse-glimmer-30b 131.072K 131.072K $0.35 $1.5 $0.04 ReasoningToolsJSON Docs ↗
NanoGPT meta/muse-glimmer-30b 131.072K 117.964K $0.35 $1.5 $0.04 ReasoningToolsJSON Docs ↗
Eden AI fireworks_ai/accounts/fireworks/models/muse-glimmer-30b 131.072K 131.072K $0.35 $1.5 $0.04 ReasoningToolsJSON Docs ↗
Fireworks AI accounts/fireworks/models/muse-glimmer-30b 131.072K 131.072K $0.35 $1.5 $0.04 ReasoningToolsJSON Docs ↗

Capability badges appear only where the provider catalog explicitly lists support. A blank cell means the source is silent, not that the feature is absent.

Listed rates

Price across providers

Input price per million tokens as published by each provider. Bars are drawn from listed rates only — no traffic weighting, since the catalog observes no requests.

Lowest input $0.2/M

Across 10 priced providers

Median input $0.325/M

Midpoint of listed rates

Highest input $0.35/M

1.7× the lowest listed rate

Output range $0.8 – $1.5

Per million output tokens

EmpirioLabs AI $0.2/MLowest
Eden AI $0.3/M
Kilo Gateway $0.3/M
OpenRouter $0.3/M
Eden AI $0.3/M
Eden AI $0.35/M
Vercel AI Gateway $0.35/M
NanoGPT $0.35/M
Eden AI $0.35/M
Fireworks AI $0.35/M
Cost calculator

Estimate a workload

$0.00
Excludes taxes, non-token charges, and tiered discounts.
Specification

Capabilities

Recorded from the source catalog and provider listings.

✓ Reasoning Yes
✓ Tool calling Yes
✓ Structured output Yes
✓ Attachments Yes
✓ Vision input Yes
✓ Open weights Yes
Creator
Meta
Model family
muse
Knowledge cutoff
2026-01-04
License
Apache 2.0
Release date
2026-08-10
Model ID
meta/muse-glimmer-30b

Weights: Hugging Face ↗

Published evaluations

Benchmarks

Every result stays attached to its source, version, metric and harness. Scores from different versions are never merged, and the profile below plots each benchmark against its own population rather than on a shared scale.

Benchmark registry
Benchmark profile
SWE-Bench ProSWE-Bench Pro GPQA DiamondGPQA Diamond OSWorld-VerifiedOSWorld-Verified Terminal-BenchTerminal-Bench 2.1 CharXiv Reasoni…CharXiv Reasoning MCP AtlasMCP Atlas This model — SWE-Bench Pro: 51.2 resolve rate (38.2th percentile) This model — GPQA Diamond: 83.5 accuracy (8.3th percentile) This model — OSWorld-Verified: 65.9 success rate (8.8th percentile) This model — Terminal-Bench 2.1: 51.7 success rate (12.5th percentile) This model — CharXiv Reasoning: 78.8 accuracy (21.4th percentile) This model — MCP Atlas: 75.5 success rate (35.7th percentile)

Each axis is this model's percentile among the 6 benchmarks it has published results for, measured against every other model with a score on that same benchmark. Percentiles are used because benchmarks do not share a scale — a 60 on one is not a 60 on another. Hover any point for the raw score.

SWE-Bench Pro resolve rate · 2026-08-10 Source ↗
SWE-Bench Pro #24 of 38
GPQA Diamond accuracy · 2026-08-10 Source ↗
GPQA Diamond #17 of 18
OSWorld-Verified success rate · 2026-08-10 Source ↗
OSWorld-Verified #16 of 17
Terminal-Bench — 2.1 success rate · 2026-08-10 Source ↗
Terminal-Bench #11 of 12
CharXiv Reasoning accuracy · 2026-08-10 Source ↗
CharXiv Reasoning #6 of 7
MCP Atlas success rate · 2026-08-10 Source ↗
MCP Atlas #5 of 7
AIME 2026 accuracy · 2026-08-10 Source ↗
94.7 1 model scored — too few for a distribution
DeepSearch QA score · 2026-08-10 Source ↗
74.6 1 model scored — too few for a distribution
SWE-Bench Verified resolve rate · 2026-08-10 Source ↗
76.0 1 model scored — too few for a distribution

Each strip shows every published score for that benchmark, with this model marked. The lighter shape behind the ticks is the density of results, and the dashed line is the median.

Cost against capability

Price and performance

Listed input price plotted against SWE-Bench Pro, the benchmark with the widest published coverage that this model appears in.

0 43 85 $1 $10 Qwen3 235B-A22B — $0.7/M, score 21.41 Qwen3-Coder 480B-A35B Instruct — $1.5/M, score 38.7 Qwen3.7 Max — $2.5/M, score 60.6 Claude Fable 5 — $10/M, score 80.3 Claude Haiku 4.5 (latest) — $1/M, score 39.45 Claude Opus 4.5 — $5/M, score 45.89 Claude Opus 4.6 — $5/M, score 51.9 Claude Opus 4.7 — $5/M, score 64.3 Claude Opus 4.8 — $5/M, score 69.2 Claude Opus 5 — $5/M, score 79.2 Claude Sonnet 4 (latest) — $2.898/M, score 42.7 Claude Sonnet 4.5 (latest) — $3/M, score 43.6 Claude Sonnet 5 — $2/M, score 63.2 Gemini 3 Flash Preview — $0.5/M, score 34.63 Gemini 3 Pro Preview — $0.57/M, score 43.3 Gemini 3.1 Pro Preview — $2/M, score 54.2 Gemini 3.5 Flash — $1.5/M, score 55.1 Gemini 3.5 Flash Lite — $0.3/M, score 54.2 LongCat-2.0 — $0.3/M, score 59.5 Llama 4 Maverick 17B Instruct — $0.14/M, score 5.24 Muse Glimmer 30B — $0.2/M, score 51.2 Muse Spark 1.1 — $1.25/M, score 61.5 MiniMax-M2.1 — $0.3/M, score 36.81 GPT-5 — $1.25/M, score 41.78 GPT-5.2 — $1.75/M, score 29.94 GPT-5.2 Codex — $0.14/M, score 41.04 GPT-5.4 — $2.5/M, score 59.1 GPT-5.4 mini — $0.75/M, score 54.4 GPT-5.4 nano — $0.2/M, score 52.4 GPT-5.5 — $5/M, score 58.6 GPT-5.6 Luna — $0.2/M, score 62.7 GPT-5.6 Sol — $4/M, score 64.6 GPT-5.6 Terra — $2/M, score 63.4 Step 3.7 Flash — $0.185/M, score 56.3 Grok 4.5 — $2/M, score 64.7 MiMo-V2.5-Pro — $0.435/M, score 57.2 GLM-4.6 — $0.6/M, score 9.67 GLM-5.2 — $1.4/M, score 62.1 Muse Glimmer 30B Input price per million tokens (log scale) Resolve rate

The stepped line is the efficient frontier: at each price, the best score available for that money or less. A model sitting on it is not being beaten by anything cheaper. Price is log-scaled because listed rates span four orders of magnitude. Only models with both a listed price and a score on this benchmark can appear.

Catalog activity

Change log

Field-level changes detected between successful source imports.

Full change log
Max Output Tokens117964 → 131072
Price Completion1.1 → 0.8
Price Prompt0.3 → 0.2
Max Output Tokens131072 → 117964
Price Completion0.8 → 1.1
Price Prompt0.2 → 0.3
Max Output Tokens117964 → 131072
Price Completion1.1 → 0.8
Price Prompt0.3 → 0.2
Max Output Tokens131072 → 117964
Provenance

Sources & verification

Every figure on this page traces back to one of these records.

Methodology
Last verified
Sep 11, 2026
Status
Source-linked
Confidence
Medium
Catalog source
Models.dev
Models.devcatalog ↗Requesty documentationprovider documentation ↗Nvidia documentationprovider documentation ↗EmpirioLabs AI documentationprovider documentation ↗Eden AI documentationprovider documentation ↗Kilo Gateway documentationprovider documentation ↗OpenRouter documentationprovider documentation ↗Vercel AI Gateway documentationprovider documentation ↗NanoGPT documentationprovider documentation ↗Fireworks AI documentationprovider documentation ↗Terminal-Benchoriginal repository ↗
Public API

Use this record

Fetch the complete source-linked model record. No key, no account, no rate-limited tier.

API documentation
Endpoint
GET https://model.kyssta.lol/api/v1/models/meta/muse-glimmer-30b
curl
curl "https://model.kyssta.lol/api/v1/models/meta/muse-glimmer-30b"
Common questions

Frequently asked questions

Answered directly from the stored record — nothing here is generated beyond the catalog's own fields.

What is Muse Glimmer 30B?

Muse Glimmer is a 30-billion-parameter open-weight multimodal model from Meta Superintelligence Labs, distilled from Muse Spark for always-on local agents, tool use, coding, and image understanding. It is published by Meta and catalogued here from Models.dev.

How much does Muse Glimmer 30B cost?

Listed input pricing starts at $0.2 per million tokens from EmpirioLabs AI, rising to $0.35 across 10 listed providers.

What is the context length of Muse Glimmer 30B?

Muse Glimmer 30B accepts up to 131.072K tokens of context and returns up to 131.072K output tokens.

Does Muse Glimmer 30B support tool calling and structured output?

Provider catalogs list support for tool calling, structured output, reasoning, and image input.

Which providers serve Muse Glimmer 30B?

9 providers list this model: Requesty, Nvidia, EmpirioLabs AI, Eden AI, Kilo Gateway, OpenRouter and 3 more.

Are the weights for Muse Glimmer 30B open?

Yes. The weights are published and downloadable from Hugging Face under the Apache 2.0 license.

When was Muse Glimmer 30B released?

The catalog records a release date of 2026-08-10, last verified Sep 11, 2026.

More models from Meta

View all →
Llama-3.2-3B

Small open Llama base model for lightweight text generation and self-hosting

131.072K ctx $0.1/M in Llama-Guard-3-8B

Llama 3.1-based safety classifier for moderating prompts and model responses

128K ctx Meta: Llama 3.2 1B Instruct

Llama 3.2 1B is a 1-billion-parameter language model focused on efficiently performing natural language tasks, such as summarization, dialogue, and multilingual text analysis. Its smaller size allows it to operate...

60K ctx $0.027/M in Meta: Llama 3.2 3B Instruct

Llama 3.2 3B is a 3-billion-parameter multilingual large language model, optimized for advanced natural language processing tasks like dialogue generation, reasoning, and summarization. Designed with the latest transformer architecture, it...

131.072K ctx $0.05/M in