---
format: "aidr-story-markdown/v1"
id: "3d25ea67182cca55639b3dd1857affe6238124be54b543bedec3084a9705570e"
canonical_url: "https://aidr.today/3d25ea67?lang=en"
title: "1.7B Coding Agent Beats GPT-4o on SWE-bench With $1,000 Training Cost"
lang: "en"
requested_lang: "en"
available_langs: ["en","vi"]
translation_fallback: null
fallback_fields: []
published_at: "2026-10-12T17:53:59.000Z"
category: "Tools"
topics: ["coding","devtools"]
source_urls: ["https://huggingnews.com/ai/17b-coding-agent-beats-gpt-4o-on-swe-bench-with-1000-training-cost-ea401955","https://marketbrief.now/ai/17b-coding-agent-beats-gpt-4o-on-swe-bench-with-1000-training-cost-ea401955"]
summary: "Researchers led by Ricardo Delgado have developed a compact AI model capable of outperforming the industry."
---

# 1\.7B Coding Agent Beats GPT\-4o on SWE\-bench With $1,000 Training Cost

> [Open the canonical story](<https://aidr.today/3d25ea67?lang=en>)

**Published:** 2026-10-12T17:53:59.000Z
**Category:** Tools
**Topics:** coding, devtools

## Summary

Researchers led by Ricardo Delgado have developed a compact AI model capable of outperforming the industry\.

## Sources

- [Story source](<https://huggingnews.com/ai/17b-coding-agent-beats-gpt-4o-on-swe-bench-with-1000-training-cost-ea401955>)
- [Story source](<https://marketbrief.now/ai/17b-coding-agent-beats-gpt-4o-on-swe-bench-with-1000-training-cost-ea401955>)

