Codex vs. DeepSeek Harness

Name: Coding Agent Index
Creator: Artificial Analysis
License: https://artificialanalysis.ai/docs/legal/Terms-of-Use.pdf

Artificial Analysis Coding Agent Index의 벤치마크 점수, 비용, 실행 시간, 토큰 사용량을 기준으로 Codex 및 DeepSeek Harness를 비교합니다.

방법론에 관한 자세한 내용은 방법론 페이지를 참조하세요.

다른 비교 살펴보기

Coding Agent Index

Artificial Analysis Coding Agent Index v2.0 · Higher is better

작업당 시간

Average agent wall time per task · Lower is better

작업당 비용

Average API cost per task (USD) · Lower is better

비교

Codex 및 DeepSeek Harness의 나란히 비교.

지표	Codex GPT-6 Astra (Max)	DeepSeek Harness DeepSeek V4.1 Flash 1017 (Max)	분석
에이전트 하네스	Codex	DeepSeek Harness
대표 모델	GPT-6 Astra (Max)	DeepSeek V4.1 Flash 1017 (Max)
Coding Agent Index	55	46	Codex의 Coding Agent Index가 DeepSeek Harness보다 높습니다
FrontierCode v1.1	49	47	Codex의 FrontierCode v1.1 점수가 DeepSeek Harness보다 높습니다
Terminal-Bench 4.0	56	31	Codex의 Terminal-Bench 4.0 점수가 DeepSeek Harness보다 높습니다
SWE-Atlas-QnA-AA	59	62	DeepSeek Harness의 SWE-Atlas-QnA-AA 점수가 Codex보다 높습니다
작업당 비용	$7.04	$0.30	DeepSeek Harness의 작업당 비용이 Codex보다 낮습니다
작업당 시간	24.8m	21.4m	DeepSeek Harness의 작업당 시간이 Codex보다 짧습니다
작업당 턴	39.4	101.5	Codex의 작업당 턴이 DeepSeek Harness보다 적습니다
작업당 토큰 사용량	3.23M	16.7M	Codex의 작업당 토큰 사용량이 DeepSeek Harness보다 적습니다
캐시 적중률	93%	97%	DeepSeek Harness의 캐시 적중률이 Codex보다 높습니다

모델 변형

Codex 및 DeepSeek Harness의 평가된 모델 변형.


Codex	GPT-6 Astra (Max)	max	55	49	56	59	$7.04	24.8m	3.23M
Codex	GPT-6.1 Sol (Max)	max	54	49	53	60	$1.50	23.7m	3.94M
Codex	GPT-6.1 Terra (Xhigh)	xhigh	48	45	46	54	$0.69	16.4m	6.89M
Codex	GPT-6.1 Terra (Max)	max	48	42	48	54	$1.24	26.1m	12.3M
Codex	DeepSeek V4.1 Flash (Max)	max	47	53	28	61	$0.39	38.1m	27.2M
Codex	GPT-6.1 Terra (High)	high	46	43	42	53	$0.54	12.7m	5.6M
Codex	GPT-6.1 Terra (Medium)	medium	42	44	35	47	$0.39	10.2m	4.08M
Codex	GPT-6.1 Terra (Non-reasoning)	non-reasoning	39	46	25	46	$0.45	9.5m	5.78M
Codex	GPT-6.1 Terra (Low)	low	35	41	25	39	$0.20	6.3m	1.89M
Codex	GPT-6 Luna (Max)	max	31	40	16	38	$0.16	20.8m	8.69M
DeepSeek Harness	DeepSeek V4.1 Flash 1017 (Max)	max	46	47	31	62	$0.30	21.4m	16.7M

성능

Artificial Analysis Coding Agent Index 전반의 성능.

Artificial Analysis Coding Agent Index

Artificial Analysis Coding Agent Index v2.0 incorporates 3 benchmarks: FrontierCode v1.1, Terminal-Bench 4.0, and SWE-Atlas-QnA-AA · Higher is better

색상 기준

The Artificial Analysis Coding Agent Index is a composite score built from FrontierCode v1.1, Terminal-Bench 4.0 및 SWE-Atlas-QnA-AA.

It is useful for quick comparison, but it should be read alongside the per-eval breakdowns. Two agents with similar index values can still have different strengths across repository tasks, terminal workflows, and rubric-based evaluations.

Artificial Analysis Coding Agent Index vs. 작업당 비용

Artificial Analysis Coding Agent Index vs. 작업당 평균 토큰당 종량제 API 비용 (USD)

색상 기준

Most attractive quadrant

Pareto line

Each point represents a coding-agent variant. Farther left means lower average cost per task, while higher on the chart means higher benchmark performance. The most efficient agents sit toward the upper-left: stronger results at lower cost.

토큰 사용량

Artificial Analysis Coding Agent Index 전반의 토큰 소비량.

작업당 토큰 사용량

작업당 평균 입력·캐시·출력 토큰

프롬프트 캐시 적중률은 제공업체 라우팅에 따라 크게 달라질 수 있으며, 이에 따라 실제 비용도 크게 달라질 수 있습니다.

Non-cached input tokens sent to the model, including prompts, instructions, tool context, and task context that were not served from prompt cache.

Artificial Analysis Coding Agent Index vs. 총 토큰

Artificial Analysis Coding Agent Index vs. 작업당 평균 총 토큰

색상 기준

Most attractive quadrant

Each point represents a coding-agent variant. Farther left means lower average total token usage per task, while higher on the chart means higher benchmark performance. Agents toward the upper-left achieve stronger results with fewer tokens.

비용

현재 토큰당 가격을 기준으로 한 Artificial Analysis Coding Agent Index의 토큰당 종량제 API 비용.

작업당 비용

Average pay-per-token API cost per task (USD) · Lower is better

색상 기준

This chart shows the average pay-per-token API cost per task across the Artificial Analysis Coding Agent Index, spanning FrontierCode v1.1, Terminal-Bench 4.0 및 SWE-Atlas-QnA-AA.

Where applicable, that cost model includes standard input pricing, discounted cached-input pricing, separate cache-write charges, and output pricing rather than treating all prompt tokens as if they were billed at the same uncached input rate.

It is intended to show pay-per-token API cost, not consumer plan pricing or the full operational cost of deploying the system in production. Infrastructure, engineering, and supervision costs are not the focus of this metric.

실행 시간

Artificial Analysis Coding Agent Index 전반의 활성 에이전트 실행 시간.

작업당 시간

Average agent wall time per task · Lower is better

색상 기준

This chart uses agent wall time: how long the agent process was actively running on each task.

It does not include environment startup, verifier or judge time, or other harness overhead, so it is a cleaner comparison of how long the agent itself was working.

Artificial Analysis Coding Agent Index vs. 실행 시간

Artificial Analysis Coding Agent Index vs. 작업당 평균 에이전트 실행 시간

색상 기준

Most attractive quadrant

Each point represents a coding-agent variant. Farther left means shorter average agent runtime per task, while higher on the chart means higher benchmark performance. Agents toward the upper-left deliver stronger results in less active agent time.

Codex vs. DeepSeek Harness

Coding Agent Index

작업당 시간

작업당 비용

비교

모델 변형

성능

Artificial Analysis Coding Agent Index

What This Metric Means

Artificial Analysis Coding Agent Index vs. 작업당 비용

How to Read This Chart

토큰 사용량

작업당 토큰 사용량

Input Tokens

Artificial Analysis Coding Agent Index vs. 총 토큰

How to Read This Chart

비용

작업당 비용

What Cost Is Measuring

실행 시간

작업당 시간

What Execution Time Is Measuring

Artificial Analysis Coding Agent Index vs. 실행 시간

How to Read This Chart