vs
0:8
overall capability wins (0 ties across 8 scored). Muse Spark 1.2 leads overall.
- Reasoning0:4
- Coding0:1
- Arena0:1
- Pricing1:2
- Speed0:2
- Specs0:2
Verdict
Muse Spark 1.2 leads coding (widest gap: +27.5 on SciCode); Muse Spark 1.2 ranks higher on LMArena (+164 Elo); Muse Spark 1.2 offers ~8.2x larger context (1.0M vs 128K).
- Muse Spark 1.2 leads coding (widest gap: +27.5 on SciCode)
- Muse Spark 1.2 ranks higher on LMArena (+164 Elo)
- Muse Spark 1.2 offers ~8.2x larger context (1.0M vs 128K)
- Muse Spark 1.2 allows ~32.0x more max output (131K vs 4K)
- Llama 3.1 405B ships open weights (Llama 3.1 Community)
- Muse Spark 1.2 is ~1.8x cheaper on a blended token basis than Llama 3.1 405B
Cost for 1M in + 250K out
Illustrative chat workload at primary-provider list prices.
- Llama 3.1 405B
- $4.38
- Muse Spark 1.2
- $2.31
Muse Spark 1.2 is about 1.9x cheaper on this workload.
Price per 1M tokens
Full comparison
Organization
Meta
Meta
Family
Llama / Muse
Llama / Muse
License
Llama 3.1 Community
Proprietary
Open weights
Yes
No
Release
Jul 23, 2024
Aug 5, 2026
Knowledge cutoff
-
-
API / provider
Together / Fireworks (ref.)
Meta
Modalities
text → text
text, image, video, audio, pdf → text
Specs
Context window
128K
1.0M
Max output
4K
131K
Parameters
405B
—
Pricing
Input $/1M
$3.50
$1.25
Output $/1M
$3.50
$4.25
Blended $/1M (3∶1)
$3.50
$2
Speed
tok/s
40
95
TTFT (s)
0.6
0.38
Reasoning
MMLU-Pro
73.3
—
GPQA Diamond
51.5
90.4
Humanity's Last Exam
4.0
45.5
AIME 2025
3.0
—
MATH-500
70.3
—
Humanity's Last Exam (with tools)
—
—
AA-Omniscience Accuracy
23.2
45.4
AA-LCR v1.1
25.3
79.0
CritPt
—
17.7
MMMU-Pro
—
—
IFBench
39.0
—
Chartography
—
—
Chartography (With Tools)
—
—
Coding
SWE-bench Verified
—
—
SWE-bench Pro
11.2
—
SWE-bench Multilingual
—
—
LiveCodeBench
27.7
—
Terminal-Bench 2.1
—
80.1
Aider Polyglot
—
—
Terminal-Bench 3
—
—
BigCodeBench
—
—
SciCode
29.9
57.4
CursorBench
—
—
SWE-Rebench
—
—
NL2Repo-Bench
—
—
DeepSWE
—
55.0
WebDev Arena
—
1,535
Terminal-Bench 4.0
—
—
CursorBench 4.0
—
—
FrontierCode v1.1 (Main)
—
—
Arena
LMArena Elo
1,335
1,500
| Metric | Llama 3.1 405B | Muse Spark 1.2 | Delta |
|---|---|---|---|
| Identity | |||
| Organization | Meta | Meta | — |
| Family | Llama / Muse | Llama / Muse | — |
| License | Llama 3.1 Community | Proprietary | — |
| Open weights | Yes | No | — |
| Release | Jul 23, 2024 | Aug 5, 2026 | — |
| Knowledge cutoff | - | - | — |
| API / provider | Together / Fireworks (ref.) | Meta | — |
| Modalities | text → text | text, image, video, audio, pdf → text | — |
| Specs | |||
| Context window | 128K | 1.0M | B +921K |
| Max output | 4K | 131K | B +127K |
| Parameters | 405B | — | — |
| Pricing | |||
| Input $/1M | $3.50 | $1.25 | B +$2.25 |
| Output $/1M | $3.50 | $4.25 | A +$0.75 |
| Blended $/1M (3∶1) | $3.50 | $2 | B +$1.50 |
| Speed | |||
| tok/s | 40 | 95 | B +55 |
| TTFT (s) | 0.6 | 0.38 | B +0.22 |
| Reasoning | |||
| MMLU-Pro | 73.3 | — | — |
| GPQA Diamond | 51.5 | 90.4 | B +38.9 pts |
| Humanity's Last Exam | 4.0 | 45.5 | B +41.5 pts |
| AIME 2025 | 3.0 | — | — |
| MATH-500 | 70.3 | — | — |
| Humanity's Last Exam (with tools) | — | — | — |
| AA-Omniscience Accuracy | 23.2 | 45.4 | B +22.2 pts |
| AA-LCR v1.1 | 25.3 | 79.0 | B +53.7 pts |
| CritPt | — | 17.7 | — |
| MMMU-Pro | — | — | — |
| IFBench | 39.0 | — | — |
| Chartography | — | — | — |
| Chartography (With Tools) | — | — | — |
| Coding | |||
| SWE-bench Verified | — | — | — |
| SWE-bench Pro | 11.2 | — | — |
| SWE-bench Multilingual | — | — | — |
| LiveCodeBench | 27.7 | — | — |
| Terminal-Bench 2.1 | — | 80.1 | — |
| Aider Polyglot | — | — | — |
| Terminal-Bench 3 | — | — | — |
| BigCodeBench | — | — | — |
| SciCode | 29.9 | 57.4 | B +27.5 pts |
| CursorBench | — | — | — |
| SWE-Rebench | — | — | — |
| NL2Repo-Bench | — | — | — |
| DeepSWE | — | 55.0 | — |
| WebDev Arena | — | 1,535 | — |
| Terminal-Bench 4.0 | — | — | — |
| CursorBench 4.0 | — | — | — |
| FrontierCode v1.1 (Main) | — | — | — |
| Arena | |||
| LMArena Elo | 1,335 | 1,500 | B +164 Elo |
Llama 3.1 405B
- Context
- 128K / 4K out
- Parameters
- 405B
- Price
- $3.50 / $3.50
- Speed
- 40 tok/s · 0.6s TTFT
- Modalities
- text → text
- License
- Llama 3.1 Community
Largest Llama 3.1 dense instruct model - still used where dense (non-MoE) weights are required.
Muse Spark 1.2
- Context
- 1.0M / 131K out
- Parameters
- —
- Price
- $1.25 / $4.25
- Speed
- 95 tok/s · 0.38s TTFT
- Modalities
- text, image, video, audio, pdf → text
- License
- Proprietary
Co-trained with Muse Code for higher first-attempt accuracy and more reliable tool calling — 1M context coding-focused upgrade; Meta Model API at $1.25/$4.25 (contributor tier $0.10/$0.20).
Benchmark charts
Winner bars are emphasized. Per-benchmark deltas sit above each chart.