<!-- mobian-agent-page publisher="dailydev" canonical="https://daily.dev/posts/openai-says-its-agents-now-finish-nearly-half-of-2-to-4-hour-research-tasks-mxr2vaoip" -->

---
title: OpenAI says its agents now finish nearly half of 2 to...
description: OpenAI's coding agents now complete 46% of two-to-four-hour research tasks alone, up from 23% in March. For four-to-eight-hour tasks, success rose from 13% to...
canonical: https://daily.dev/posts/openai-says-its-agents-now-finish-nearly-half-of-2-to-4-hour-research-tasks-mxr2vaoip
twitter:card: summary_large_image
twitter:site: @dailydotdev
og:type: website
og:site_name: daily.dev
og:title: OpenAI says its agents now finish nearly half of 2 to 4-hour research tasks | daily.dev
og:description: OpenAI's coding agents now complete 46% of two-to-four-hour research tasks alone, up from 23% in March. For four-to-eight-hour tasks, success rose from 13% to...
og:url: https://daily.dev/posts/openai-says-its-agents-now-finish-nearly-half-of-2-to-4-hour-research-tasks-mxr2vaoip
og:image: https://api.daily.dev/og/posts/mxR2vaOIP.png
og:image:alt: OpenAI says its agents now finish nearly half of 2 to 4-hour research tasks
og:image:width: 1200
og:image:height: 630
og:locale: en
---

> ## Documentation Index
> Fetch the complete documentation index at: https://daily.dev/llms.txt
> Use this file to discover all available pages before exploring further.

# OpenAI says its agents now finish nearly half of 2 to 4-hour research tasks

**[The Next Web](https://daily.dev/sources/tnw)** · 4 min read · 0 upvotes · 0 comments

## Summary

OpenAI's coding agents now complete 46% of two-to-four-hour research tasks alone, up from 23% in March. For four-to-eight-hour tasks, success rose from 13% to 35%. The median researcher used about $2,400 of agent time daily in October, up 62% from September, and the 90th percentile hit $23,000. Agent use per researcher rose elevenfold to 5.7 'agent workdays' a day. Agents succeeded alone 79% of the time when monitoring training runs, but 25% when writing code for inference and serving systems. In a survey of 101 research staff, the median answer was 40% of team headcount to keep the agents for three months.

## Full article

daily.dev links to this article rather than hosting it. Read it at the original source: <https://thenextweb.com/news/openai-research-agents-46-percent-tasks-2400-a-day-jagged-frontier>

## Questions this post answers

### What percentage of 2 to 4-hour research tasks can OpenAI's coding agents complete without human help?

OpenAI's coding agents complete 46% of research tasks that would take a person two to four hours, with no human intervention, up from 23% in March. For longer four-to-eight-hour tasks, success rose from 13% to 35% over the same period, according to OpenAI's own internal reporting on AI-accelerated research.

_Developers weighing how far coding agents can run unsupervised can track these benchmarks on daily.dev._

### How much does the median OpenAI researcher spend on AI agent usage per day?

The median OpenAI researcher used about $2,400 of agent time per day in October, measured at API list prices, up 62% from $1,500 in September. At the 90th percentile, daily usage reached roughly $23,000, more than double the prior month's $11,000, reflecting rapidly rising internal reliance on agents.

_Teams budgeting for agentic coding workflows can follow real-world cost data like this on daily.dev._

### What tasks are AI coding agents still bad at compared to monitoring or answering questions?

Coding agents at OpenAI succeed only 25% of the time writing code for inference and serving systems, and 29% of the time building evaluation and training environments, compared to 79% success monitoring training runs and 72% answering technical questions. The company calls this unevenness the 'jagged frontier' of agent capability.

_Engineers deciding which coding tasks to delegate to agents can compare these gaps on daily.dev._

---

Tags: [#ai](https://daily.dev/tags/ai), [#ai-agents](https://daily.dev/tags/ai-agents), [#openai](https://daily.dev/tags/openai), [#ai-safety](https://daily.dev/tags/ai-safety), [#openai-codex](https://daily.dev/tags/openai-codex)

[View this post on daily.dev](https://daily.dev/posts/openai-says-its-agents-now-finish-nearly-half-of-2-to-4-hour-research-tasks-mxr2vaoip)

```json
{"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]}
{"@context":"https://schema.org","@type":"TechArticle","headline":"OpenAI says its agents now finish nearly half of 2 to 4-hour research tasks","url":"https://daily.dev/posts/openai-says-its-agents-now-finish-nearly-half-of-2-to-4-hour-research-tasks-mxr2vaoip","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/openai-says-its-agents-now-finish-nearly-half-of-2-to-4-hour-research-tasks-mxr2vaoip"},"datePublished":"2026-11-21T17:59:16.322Z","dateModified":"2026-11-21T18:00:13.114Z","description":"OpenAI's coding agents now complete 46% of two-to-four-hour research tasks alone, up from 23% in March. For four-to-eight-hour tasks, success rose from 13% to...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/0b534a6f71eb7c59d1961e25fb253f6f?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/0b534a6f71eb7c59d1961e25fb253f6f?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"The Next Web","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"The Next Web","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/tnw","url":"https://daily.dev/sources/tnw"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/openai-says-its-agents-now-finish-nearly-half-of-2-to-4-hour-research-tasks-mxr2vaoip","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"ai,ai-agents,openai,ai-safety,openai-codex","timeRequired":"PT4M"}
{"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"The Next Web","item":"https://daily.dev/sources/tnw"},{"@type":"ListItem","position":3,"name":"OpenAI says its agents now finish nearly half of 2 to 4-hour research tasks"}]}
{"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/openai-says-its-agents-now-finish-nearly-half-of-2-to-4-hour-research-tasks-mxr2vaoip#faq","mainEntity":[{"@type":"Question","name":"What percentage of 2 to 4-hour research tasks can OpenAI's coding agents complete without human help?","acceptedAnswer":{"@type":"Answer","text":"OpenAI's coding agents complete 46% of research tasks that would take a person two to four hours, with no human intervention, up from 23% in March. For longer four-to-eight-hour tasks, success rose from 13% to 35% over the same period, according to OpenAI's own internal reporting on AI-accelerated research. Developers weighing how far coding agents can run unsupervised can track these benchmarks on daily.dev."}},{"@type":"Question","name":"How much does the median OpenAI researcher spend on AI agent usage per day?","acceptedAnswer":{"@type":"Answer","text":"The median OpenAI researcher used about $2,400 of agent time per day in October, measured at API list prices, up 62% from $1,500 in September. At the 90th percentile, daily usage reached roughly $23,000, more than double the prior month's $11,000, reflecting rapidly rising internal reliance on agents. Teams budgeting for agentic coding workflows can follow real-world cost data like this on daily.dev."}},{"@type":"Question","name":"What tasks are AI coding agents still bad at compared to monitoring or answering questions?","acceptedAnswer":{"@type":"Answer","text":"Coding agents at OpenAI succeed only 25% of the time writing code for inference and serving systems, and 29% of the time building evaluation and training environments, compared to 79% success monitoring training runs and 72% answering technical questions. The company calls this unevenness the 'jagged frontier' of agent capability. Engineers deciding which coding tasks to delegate to agents can compare these gaps on daily.dev."}}]}
```

