Inference
EN

GPT-5.2-Codex vs Muse Spark

Change models
GPT-5.2-CodexOpenAI
Muse SparkMeta
IndexAll categories, out of 100.
59.9 ±4.2 best
59.8 ±3.5
GapIndex points below the top.
Top
−0.1 · no clear gap
CoverageIndex weight that has results.
95%
95%
SpeedOutput tokens per second.
44/s
—
EloLMArena rating and rank.
—
1474
ReleasedMonth of release.
2025-12
2026-04
Input / 1MUS dollars per 1M input tokens.
$1.75
—
Output / 1MUS dollars per 1M output tokens.
$14.00
—
1K chats2K tokens in, 500 out, each.
$10.50
—
1K documents20K tokens in, 1K out, each.
$49.00
—
1K agent tasks200K tokens in, 20K out, each.
$630.00
—
ContextMaximum tokens in one request.
400K best
262K
AgenticMulti-step tasks with tools.
58.1 best
55.9
CodingCode writing and repair.
61.4 best
53.3
ReasoningLogic problems and puzzles.
60.0 best
59.8
MultimodalTasks with images and text.
54.5
63.0 best
KnowledgeFacts and expert knowledge.
60.1
67.8 best
MultilingualTasks in many languages.
—
—
InstructionTasks with strict prompt rules.
61.3 best
60.0
MathMath problems.
63.5 best
60.2

Benchmarks

10 shared
Only GPT-5.2-Codex: 12Only Muse Spark: 37
GPT-5.2-Codex
Muse Spark
Artificial Analysis Humanity's Last ExamKnowledge
35.7%59 / 222
40.7% best45 / 222
Artificial Analysis Long Context ReasoningReasoning
82.3% best27 / 217
78.0%70 / 217
Critical Physics TasksReasoning
8.7%73 / 214
11.3% best60 / 214
Artificial Analysis GPQA DiamondKnowledge
89.9% best37 / 199
88.4%51 / 199
Artificial Analysis Omniscience AccuracyKnowledge
41.1%45 / 164
49.6% best21 / 164
τ²-Bench Tool-Agent-User EvaluationAgentic
92.1% best35 / 149
91.5%38 / 149
Artificial Analysis IFBenchInstruction
77.6% best9 / 146
75.9%16 / 146
Vibe Code Bench v1.1Coding
37.9% best76 / 123
19.7%92 / 123
Artificial Analysis MMMU-ProMultimodal
76.3%46 / 113
80.5% best18 / 113
SWE-benchCoding
72.4%55 / 87
74.4% best48 / 87