GPT-5.2-Codex

openai/gpt-5.2-codex-20260114

OpenAI's GPT-5.2-Codex processes text and image inputs to generate text output, with a context length of 400 000 tokens. The model supports tool usage and reasoning, is not released with open weights, and costs $1.75 per M input tokens and $14 per M output tokens. It was released on 2026‑01‑14.

  • tool use
  • structured outputs
  • reasoning

At a glance

provideropenai
releasedJan 15, 2025
context400K
input / M$1.75

Specifications

What the API gives you

Context window
400K tokens
Max output
128K tokens
Input price
$1.75 / 1M tokens
Output price
$14 / 1M tokens
Cache read
$0.17 / 1M tokens
Modalities
text, image → text
Knowledge cutoff
Released
Jan 15, 2025

Measured evidence

Public benchmark record

Design Arena Web-app Agent
1109 Elo2026-07-28
Design Arena Web-app Agent
1109 Elo2026-07-27
Design Arena Web-app Agent
1106 Elo2026-07-30
Design Arena Web-app Agent
1106 Elo2026-07-29
Design Arena Web-app Agent
1106 Elo2026-07-31
Design Arena Web-app Agent
1106 Elo2026-08-02
Design Arena Web-app Agent
1106 Elo2026-08-01
Design Arena Web-app Agent
1099 Elo2026-08-05
Design Arena Web-app Agent
1094 Elo2026-08-09
Design Arena Web-app Agent
1106 Elo2026-08-03
Design Arena Web-app Agent
1107 Elo2026-08-04
Design Arena Web-app Agent
1089 Elo2026-08-14
Design Arena Web-app Agent
1100 Elo2026-08-06
Design Arena Web-app Agent
1091 Elo2026-08-10
Design Arena Web-app Agent
1100 Elo2026-08-07
Design Arena Web-app Agent
1100 Elo2026-08-08
Design Arena Web-app Agent
1088 Elo2026-08-18
Design Arena Web-app Agent
1091 Elo2026-08-11
Design Arena Web-app Agent
1089 Elo2026-08-15
Design Arena Web-app Agent
1092 Elo2026-08-12
Design Arena Web-app Agent
1090 Elo2026-08-13
Design Arena Web-app Agent
1088 Elo2026-08-19
Design Arena Web-app Agent
1082 Elo2026-08-27
Design Arena Web-app Agent
1088 Elo2026-08-20
Design Arena Web-app Agent
1091 Elo2026-08-16
Design Arena Web-app Agent
1092 Elo2026-08-17
Design Arena Web-app Agent
1081 Elo2026-09-05
Design Arena Web-app Agent
1088 Elo2026-08-21
Design Arena Web-app Agent
1088 Elo2026-08-22
Design Arena Web-app Agent
1088 Elo2026-08-23
Design Arena Web-app Agent
1082 Elo2026-08-28
Design Arena Web-app Agent
1083 Elo2026-08-24
Design Arena Web-app Agent
1084 Elo2026-08-25
Design Arena Web-app Agent
1084 Elo2026-08-26
Design Arena Web-app Agent
1081 Elo2026-09-06
Design Arena Web-app Agent
1082 Elo2026-08-29
Design Arena Web-app Agent
1082 Elo2026-08-30
Design Arena Web-app Agent
1082 Elo2026-08-31
Design Arena Web-app Agent
1081 Elo2026-09-01
Design Arena Web-app Agent
1081 Elo2026-09-02
Design Arena Web-app Agent
1081 Elo2026-09-03
Design Arena Web-app Agent
1081 Elo2026-09-04
LiveBench Coding
83.6%2026-06-25
LiveBench Language
73.7%2026-06-25
LiveBench Math
88.8%2026-06-25
LiveBench Reasoning
77.7%2026-06-25
LMArena WebDev
1334 Elo2026-07-16
LMArena WebDev
1334 Elo2026-07-21
LMArena WebDev
1334 Elo2026-07-20
LMArena WebDev
1338 Elo2026-07-27
LMArena WebDev
1338 Elo2026-07-24
LMArena WebDev
1338 Elo2026-07-28
LMArena WebDev
1338 Elo2026-07-30
LMArena WebDev
1338 Elo2026-08-01
LMArena WebDev
1338 Elo2026-07-31
LMArena WebDev
1338 Elo2026-08-06
LMArena WebDev
1338 Elo2026-08-02
LMArena WebDev
1338 Elo2026-08-12
LMArena WebDev
1338 Elo2026-08-10
LMArena WebDev
1338 Elo2026-08-14
LMArena WebDev
1339 Elo2026-09-01
LMArena WebDev
1338 Elo2026-09-05
LMArena WebDev
1338 Elo2026-08-21
LMArena WebDev
1338 Elo2026-08-30
SWE-rebench
51.3% ± 1.22026-03-01tools

LMArena data licensed CC-BY-4.0. Aider (Apache-2.0) and SWE-bench (MIT) are open. Design Arena uses blind human preference. Submitted scaffold plus model result. Compare the harness and effort settings, not only the model name.

Task evidence

How broad is the support?

Each status is exact to this model version and the selected task. Official release claims remain provisional.

  • Best LLM for Code CompletionEstablished evidence

    81% of intended weighted task evidence across 2 independent sources. 4 of 4 configured evaluation feeds currently have results.

  • 70% of intended weighted task evidence across 3 independent sources. 6 of 6 configured evaluation feeds currently have results.

  • Best Local LLM for CodingEstablished evidence

    64% of intended weighted task evidence across 3 independent sources. 5 of 5 configured evaluation feeds currently have results.

  • Best LLM for CodingEstablished evidence

    60% of intended weighted task evidence across 3 independent sources. 8 of 8 configured evaluation feeds currently have results.

  • Best LLM for Competition MathProvisional evidence

    65% of intended weighted task evidence across 1 independent source; broader independent confirmation is still missing from LMArena Instruction Following or LMArena Math. 5 of 5 configured evaluation feeds currently have results.

  • Best LLM for Math & ReasoningProvisional evidence

    65% of intended weighted task evidence across 1 independent source; broader independent confirmation is still missing from LMArena Instruction Following or LMArena Math. 5 of 5 configured evaluation feeds currently have results.

  • 60% of intended weighted task evidence across 1 independent source; broader independent confirmation is still missing from LMArena Instruction Following or LMArena Math. 5 of 5 configured evaluation feeds currently have results.

  • Best LLM for Agentic CodingProvisional evidence

    55% of intended weighted task evidence across 3 independent sources; broader independent confirmation is still missing from Frontier-Bench or LMArena Agent. 8 of 8 configured evaluation feeds currently have results.

  • 55% of intended weighted task evidence across 1 independent source; broader independent confirmation is still missing from LMArena Instruction Following or LMArena Text. 5 of 5 configured evaluation feeds currently have results.

  • Best LLM for Creative WritingProvisional evidence

    50% of intended weighted task evidence across 1 independent source; broader independent confirmation is still missing from LMArena Creative Writing or LMArena Text. 5 of 5 configured evaluation feeds currently have results.

  • Best LLM for WritingProvisional evidence

    50% of intended weighted task evidence across 1 independent source; broader independent confirmation is still missing from LMArena Creative Writing or LMArena Text. 5 of 5 configured evaluation feeds currently have results.

  • Best LLM for Fast SummariesProvisional evidence

    50% of intended weighted task evidence across 1 independent source; broader independent confirmation is still missing from LMArena Text or Route reliability. 3 of 4 configured evaluation feeds currently have results.

  • Best Value LLMProvisional evidence

    43% of intended weighted task evidence across 1 independent source; broader independent confirmation is still missing from LMArena Text or Route reliability. 3 of 4 configured evaluation feeds currently have results.

  • Best Cheap LLM for High VolumeProvisional evidence

    42% of intended weighted task evidence across 1 independent source; broader independent confirmation is still missing from LMArena Text or Route reliability. 3 of 4 configured evaluation feeds currently have results.

  • Cheapest Capable LLMProvisional evidence

    42% of intended weighted task evidence across 1 independent source; broader independent confirmation is still missing from LMArena Text or Route reliability. 3 of 4 configured evaluation feeds currently have results.

  • 42% of intended weighted task evidence across 1 independent source; broader independent confirmation is still missing from LMArena Text or Route reliability. 3 of 4 configured evaluation feeds currently have results.

  • Best LLM for Autonomous AgentsProvisional evidence

    41% of intended weighted task evidence across 2 independent sources; broader independent confirmation is still missing from Frontier-Bench or LMArena Agent. 7 of 7 configured evaluation feeds currently have results.

  • 40% of intended weighted task evidence across 1 independent source; broader independent confirmation is still missing from LMArena Text or WMT translation. 4 of 5 configured evaluation feeds currently have results.

  • 40% of intended weighted task evidence across 1 independent source; broader independent confirmation is still missing from LMArena Text or WMT translation. 4 of 5 configured evaluation feeds currently have results.

  • Best LLM for TranslationProvisional evidence

    40% of intended weighted task evidence across 1 independent source; broader independent confirmation is still missing from LMArena Text or WMT translation. 4 of 5 configured evaluation feeds currently have results.

  • Best LLM for Business WritingProvisional evidence

    40% of intended weighted task evidence across 1 independent source; broader independent confirmation is still missing from LMArena Creative Writing or LMArena Instruction Following. 5 of 5 configured evaluation feeds currently have results.

  • Best LLM for Marketing CopyProvisional evidence

    40% of intended weighted task evidence across 1 independent source; broader independent confirmation is still missing from LMArena Creative Writing or LMArena Instruction Following. 5 of 5 configured evaluation feeds currently have results.

  • Best LLM for Chat & RoleplayProvisional evidence

    40% of intended weighted task evidence across 1 independent source; broader independent confirmation is still missing from LMArena Creative Writing or LMArena Text. 5 of 5 configured evaluation feeds currently have results.

  • Best LLM for Frontend & UIProvisional evidence

    40% of intended weighted task evidence across 2 independent sources; broader independent confirmation is still missing from Design Arena Coding or Design Arena UI. 5 of 5 configured evaluation feeds currently have results.

  • 40% of intended weighted task evidence across 1 independent source; broader independent confirmation is still missing from LMArena Creative Writing or LMArena Text. 5 of 5 configured evaluation feeds currently have results.

  • 31% of intended weighted task evidence across 1 independent source; broader independent confirmation is still missing from Route reliability or Structured-output eval. 2 of 4 configured evaluation feeds currently have results.

  • 30% of intended weighted task evidence across 1 independent source; broader independent confirmation is still missing from FACTS Grounding or LMArena Document. 4 of 6 configured evaluation feeds currently have results.

  • Best LLM for SummarizationProvisional evidence

    30% of intended weighted task evidence across 1 independent source; broader independent confirmation is still missing from FACTS Grounding or LMArena Document. 4 of 6 configured evaluation feeds currently have results.

  • Best LLM for AI AgentsProvisional evidence

    30% of intended weighted task evidence across 2 independent sources; broader independent confirmation is still missing from Berkeley Function Calling or LMArena Agent. 7 of 7 configured evaluation feeds currently have results.

  • 28% of intended weighted task evidence across 1 independent source; broader independent confirmation is still missing from Berkeley Function Calling or LMArena Text. 4 of 5 configured evaluation feeds currently have results.

  • 28% of intended weighted task evidence across 1 independent source; broader independent confirmation is still missing from Berkeley Function Calling or LMArena Text. 4 of 5 configured evaluation feeds currently have results.

  • 20% of intended weighted task evidence across 1 independent source; broader independent confirmation is still missing from LMArena Vision or VLMEvalKit tasks. 3 of 4 configured evaluation feeds currently have results.

  • Best LLM for Hermes AgentProvisional evidence

    18% of intended weighted task evidence across 1 independent source; broader independent confirmation is still missing from Berkeley Function Calling or LMArena Agent. 5 of 5 configured evaluation feeds currently have results.

  • Best LLM for OpenHandsProvisional evidence

    18% of intended weighted task evidence across 1 independent source; broader independent confirmation is still missing from Berkeley Function Calling or LMArena Agent. 5 of 5 configured evaluation feeds currently have results.

  • Best LLM for GooseProvisional evidence

    18% of intended weighted task evidence across 1 independent source; broader independent confirmation is still missing from Berkeley Function Calling or LMArena Agent. 5 of 5 configured evaluation feeds currently have results.

  • Best LLM for OpenClawProvisional evidence

    18% of intended weighted task evidence across 1 independent source; broader independent confirmation is still missing from Berkeley Function Calling or LMArena Agent. 5 of 5 configured evaluation feeds currently have results.

  • 10% of intended weighted task evidence across 0 independent sources; broader independent confirmation is still missing from Berkeley Function Calling or LMArena Agent. 4 of 5 configured evaluation feeds currently have results.

  • 10% of intended weighted task evidence across 0 independent sources; broader independent confirmation is still missing from Berkeley Function Calling or LMArena Agent. 4 of 4 configured evaluation feeds currently have results.

  • Best LLM for OCR & DocumentsProvisional evidence

    10% of intended weighted task evidence across 0 independent sources; broader independent confirmation is still missing from LMArena Document or LMArena Vision. 3 of 5 configured evaluation feeds currently have results.

  • 10% of intended weighted task evidence across 0 independent sources; broader independent confirmation is still missing from LMArena Document or LMArena Vision. 3 of 5 configured evaluation feeds currently have results.

  • Best LLM for Grounded RAGProvisional evidence

    10% of intended weighted task evidence across 0 independent sources; broader independent confirmation is still missing from CRAG or FACTS Grounding. 2 of 5 configured evaluation feeds currently have results.

  • Best LLM for Charts & DiagramsProvisional evidence

    10% of intended weighted task evidence across 0 independent sources; broader independent confirmation is still missing from LMArena Document or LMArena Vision. 3 of 5 configured evaluation feeds currently have results.

  • Best LLM for Document ParsingProvisional evidence

    10% of intended weighted task evidence across 0 independent sources; broader independent confirmation is still missing from LMArena Document or LMArena Vision. 3 of 5 configured evaluation feeds currently have results.

  • Best LLM for Massive ContextProvisional evidence

    6% of intended weighted task evidence across 0 independent sources; broader independent confirmation is still missing from LMArena Document or LMArena Long Query. 3 of 4 configured evaluation feeds currently have results.

  • 6% of intended weighted task evidence across 0 independent sources; broader independent confirmation is still missing from FACTS Grounding or LMArena Document. 4 of 6 configured evaluation feeds currently have results.

Recommendation coverage

Where this model appears

Not currently included in a published recommendation. The task evidence panel above says exactly what is confirmed and what is still missing.

Catalog data via OpenRouter and models.dev. Last checked within the past 6 hours.