vs
2:10
overall capability wins (0 ties across 12 scored). Muse Glimmer leads overall.
- Reasoning0:5
- Coding0:3
- Arena0:1
- Tool use0:1
- Composite indices0:2
- Specs2:0
Verdict
Muse Glimmer leads coding (widest gap: +66.9 on SWE-bench); Muse Glimmer ranks higher on LMArena (+106 Elo); Llama 4 Scout offers ~76.3x larger context (10M vs 131K).
- Muse Glimmer leads coding (widest gap: +66.9 on SWE-bench)
- Muse Glimmer ranks higher on LMArena (+106 Elo)
- Llama 4 Scout offers ~76.3x larger context (10M vs 131K)
Cost for 1M in + 250K out
Illustrative chat workload at primary-provider list prices.
- Llama 4 Scout
- $0.33
- Muse Glimmer
- -
Price per 1M tokens
Full comparison
Organization
Meta
Meta
Family
Llama / Muse
Llama / Muse
License
Llama 4 Community
Apache 2.0
Open weights
Yes
Yes
Release
Apr 5, 2025
Aug 10, 2026
Knowledge cutoff
-
2026-01-04
API / provider
Together / Fireworks (ref.)
No primary API price listed
Modalities
text, image → text
text, image → text
Specs
Context window
10M
131K
Max output
33K
-
Parameters
109B (17B act.)
30B
Pricing
Input $/1M
$0.18
—
Output $/1M
$0.59
—
Blended $/1M (3∶1)
$0.28
-
Speed
tok/s
110
-
TTFT (s)
0.3
-
Reasoning
MMLU-Pro
74.3
—
GPQA Diamond
58.7
83.5
Humanity's Last Exam
3.8
22.0
AIME 2025
14.0
—
MATH-500
84.4
—
Humanity's Last Exam (with tools)
—
—
AA-Omniscience Accuracy
15.2
27.0
AA-LCR v1.1
27.7
83.3
CritPt
—
2.6
MMMU-Pro
52.9
74.3
IFBench
39.5
—
Chartography
—
—
Chartography (With Tools)
—
—
Coding
SWE-bench Verified
9.1
76.0
SWE-bench Pro
—
51.2
SWE-bench Multilingual
—
—
LiveCodeBench
32.8
—
Terminal-Bench 2.1
3.7
51.7
Aider Polyglot
—
—
Terminal-Bench 3
—
—
BigCodeBench
—
—
SciCode
21.3
44.9
CursorBench
—
—
SWE-Rebench
6.6
—
NL2Repo-Bench
—
—
DeepSWE
—
—
WebDev Arena
—
1,360
Terminal-Bench 4.0
—
—
CursorBench 4.0
—
—
FrontierCode v1.1 (Main)
—
—
Arena
LMArena Elo
1,322
1,427
| Metric | Llama 4 Scout | Muse Glimmer | Delta |
|---|---|---|---|
| Identity | |||
| Organization | Meta | Meta | — |
| Family | Llama / Muse | Llama / Muse | — |
| License | Llama 4 Community | Apache 2.0 | — |
| Open weights | Yes | Yes | — |
| Release | Apr 5, 2025 | Aug 10, 2026 | — |
| Knowledge cutoff | - | 2026-01-04 | — |
| API / provider | Together / Fireworks (ref.) | No primary API price listed | — |
| Modalities | text, image → text | text, image → text | — |
| Specs | |||
| Context window | 10M | 131K | A +9.9M |
| Max output | 33K | - | — |
| Parameters | 109B (17B act.) | 30B | A +79 |
| Pricing | |||
| Input $/1M | $0.18 | — | — |
| Output $/1M | $0.59 | — | — |
| Blended $/1M (3∶1) | $0.28 | - | — |
| Speed | |||
| tok/s | 110 | - | — |
| TTFT (s) | 0.3 | - | — |
| Reasoning | |||
| MMLU-Pro | 74.3 | — | — |
| GPQA Diamond | 58.7 | 83.5 | B +24.8 pts |
| Humanity's Last Exam | 3.8 | 22.0 | B +18.2 pts |
| AIME 2025 | 14.0 | — | — |
| MATH-500 | 84.4 | — | — |
| Humanity's Last Exam (with tools) | — | — | — |
| AA-Omniscience Accuracy | 15.2 | 27.0 | B +11.8 pts |
| AA-LCR v1.1 | 27.7 | 83.3 | B +55.6 pts |
| CritPt | — | 2.6 | — |
| MMMU-Pro | 52.9 | 74.3 | B +21.4 pts |
| IFBench | 39.5 | — | — |
| Chartography | — | — | — |
| Chartography (With Tools) | — | — | — |
| Coding | |||
| SWE-bench Verified | 9.1 | 76.0 | B +66.9 pts |
| SWE-bench Pro | — | 51.2 | — |
| SWE-bench Multilingual | — | — | — |
| LiveCodeBench | 32.8 | — | — |
| Terminal-Bench 2.1 | 3.7 | 51.7 | B +48.0 pts |
| Aider Polyglot | — | — | — |
| Terminal-Bench 3 | — | — | — |
| BigCodeBench | — | — | — |
| SciCode | 21.3 | 44.9 | B +23.6 pts |
| CursorBench | — | — | — |
| SWE-Rebench | 6.6 | — | — |
| NL2Repo-Bench | — | — | — |
| DeepSWE | — | — | — |
| WebDev Arena | — | 1,360 | — |
| Terminal-Bench 4.0 | — | — | — |
| CursorBench 4.0 | — | — | — |
| FrontierCode v1.1 (Main) | — | — | — |
| Arena | |||
| LMArena Elo | 1,322 | 1,427 | B +106 Elo |
Llama 4 Scout
- Context
- 10M / 33K out
- Parameters
- 109B (17B act.)
- Price
- $0.18 / $0.59
- Speed
- 110 tok/s · 0.3s TTFT
- Modalities
- text, image → text
- License
- Llama 4 Community
Llama 4 variant with an extreme 10M context window for long-document retrieval workloads.
Muse Glimmer
- Context
- 131K
- Parameters
- 30B
- Price
- —
- Speed
- —
- Modalities
- text, image → text
- License
- Apache 2.0
Meta's first open-weight model from Superintelligence Labs — 30B dense multimodal agentic model distilled from Muse Spark; Apache 2.0 with 128k context, built to run local agents on a single consumer GPU.
Benchmark charts
Winner bars are emphasized. Per-benchmark deltas sit above each chart.