Inference
EN

GPT-5.2 vs GPT-5.2-Codex

Change models
GPT-5.2OpenAI
GPT-5.2-CodexOpenAI
IndexAll categories, out of 100.
59.0 ±2.3
59.9 ±4.2 best
GapIndex points below the top.
−0.9 · no clear gap
Top
CoverageIndex weight that has results.
100% best
95%
SpeedOutput tokens per second.
45/s best
44/s
EloLMArena rating and rank.
1412
—
ReleasedMonth of release.
2025-12
2025-12
Input / 1MUS dollars per 1M input tokens.
$1.75
$1.75
Output / 1MUS dollars per 1M output tokens.
$14.00
$14.00
Batch output / 1MBatch price per 1M output tokens.
$7.00
—
1K chats2K tokens in, 500 out, each.
$10.50
$10.50
1K documents20K tokens in, 1K out, each.
$49.00
$49.00
1K agent tasks200K tokens in, 20K out, each.
$630.00
$630.00
ContextMaximum tokens in one request.
400K
400K
AgenticMulti-step tasks with tools.
55.3
58.1 best
CodingCode writing and repair.
61.0
61.4 best
ReasoningLogic problems and puzzles.
63.8 best
60.0
MultimodalTasks with images and text.
51.5
54.5 best
KnowledgeFacts and expert knowledge.
60.6 best
60.1
MultilingualTasks in many languages.
51.7
—
InstructionTasks with strict prompt rules.
55.5
61.3 best
MathMath problems.
62.6
63.5 best

Benchmarks

21 shared
Only GPT-5.2: 78Only GPT-5.2-Codex: 1
GPT-5.2
GPT-5.2-Codex
Artificial Analysis Humanity's Last ExamKnowledge
37.7% best56 / 222
35.7%59 / 222
Artificial Analysis Long Context ReasoningReasoning
82.7% best23 / 217
82.3%27 / 217
Critical Physics TasksReasoning
11.6% best59 / 214
8.7%73 / 214
Artificial Analysis GPQA DiamondKnowledge
90.3% best34 / 199
89.9%37 / 199
Artificial Analysis Omniscience AccuracyKnowledge
44.3% best40 / 164
41.1%45 / 164
τ²-Bench Tool-Agent-User EvaluationAgentic
84.8%58 / 149
92.1% best35 / 149
Artificial Analysis IFBenchInstruction
75.4%23 / 146
77.6% best9 / 146
LiveCodeBenchCoding
85.4%32 / 142
88.0% best9 / 142
Vibe Code Bench v1.1Coding
53.5% best63 / 123
37.9%76 / 123
SWE-benchCoding
75.8% best43 / 87
72.4%55 / 87
LiveBench Data AnalysisReasoning
78.2%26 / 80
78.2%26 / 80
LiveBench ReasoningReasoning
83.2% best47 / 80
77.7%64 / 80
LiveBench Agentic CodingAgentic
50.3% best42 / 66
49.4%43 / 66
LiveBench CodingCoding
76.1%44 / 66
83.6% best6 / 66
LiveBench Instruction FollowingInstruction
61.8%54 / 66
66.4% best38 / 66
LiveBench LanguageKnowledge
79.8% best30 / 66
73.7%53 / 66
LiveBench MathematicsMath
93.2% best21 / 66
88.8%38 / 66
SWE-bench Verified · SWE-bench/experimentsCoding
72.8%7 / 65
72.8%7 / 65
Gert Labs Composite Game BenchmarkAgentic
46.5%28 / 52
51.8% best18 / 52
JobBenchAgentic
34.3% best22 / 36
26.0%28 / 36