conv.

All stories
AIQuiet 3d · day 10

SoL-Pi: Researchers Cut Coding Agent Costs by a Third

MIT, NVIDIA, and NTU researchers report token-efficient harness design cuts API costs by ~33% while matching performance.

What to know

  • SoL-Pi identifies four reusable harness mechanisms that cut token traffic by ~45% and API costs by ~33% while maintaining performance parity with baseline agents.
  • The finding that harness design—not just model capability—significantly impacts coding agent efficiency has sparked community interest in component-level benchmarking.
  • Researchers and practitioners are debating terminology ('harness' as enabler vs. guardrail), which models belong in benchmarks, and whether lab results transfer to production use cases.

The dispute Whether the research applies meaningfully to production systems and current frontier models, and what 'harness' actually means in different contexts. · positions read across 12 posts and comments

many voices

Harness design legitimately matters and needs more rigorous, component-level study.

  • “we definitely need more principled studies on the role of harnesses. I'd also say that there aren't too many benchmarks where the more complicated harnesses consistently outperform extremely simple agents.”

    lieret · Hacker News ↗
some voices

The research is incomplete without testing on frontier open-source and Chinese models.

  • “This is done on Nemotron models + mistral, so its not very relevant to the current frontier of cheap chinese models + big models from Claude/GPT. Big miss not having qwen or deepseek in this research.”

    vblanco · Hacker News ↗
some voices

Lab benchmarks may not translate to real-world performance differences.

  • “I ran mouse on the Frontier Harness benchmark and scored the highest pass rate, however I'm not convinced the results there actually translate to meaning the 'best' harness in practice.”

    Aeroi · Hacker News ↗

“Token efficiency therefore becomes important for scaling recursive self-improvement.”

SoL-Pi authors (Haozhe Liu et al.), Researchers · arXiv ↗ · Sep 17

Haozhe Liu Lead researcher, SoL-Pi studyRun-Ze Fan Lead researcher, harness components studyNVIDIA, NTU, MIT Collaborating institutions

SoL-Pi: Researchers Cut Coding Agent Costs by a Third
academy.dair.ai

How it unfolded 3 developments, newest first · click a bar or a number to jump articlespostscomments

Peak 9 pieces in 3h at Sep 18, 8 AM; 32 pieces over 10 days (8 articles · 12 posts · 12 comments) Sep 17, 8 AM — 1 piece · 1 post — Hacker News 1Sep 17, 11 AM — quietSep 17, 2 PM — quietSep 17, 5 PM — quietSep 17, 8 PM — quietSep 17, 11 PM — 4 pieces · 4 articles — Newswires 4Sep 18, 2 AM — quietSep 18, 5 AM — quietSep 18, 8 AM — 9 pieces · 2 articles · 1 post · 6 comments — Hacker News 7, Newswires 2Sep 18, 11 AM — 4 pieces · 4 comments — Hacker News 4Sep 18, 2 PM — 2 pieces · 2 comments — Hacker News 2Sep 18, 5 PM — 2 pieces · 2 posts — Hacker News 1, Mastodon 1Sep 18, 8 PM — quietSep 18, 11 PM — quietSep 19, 2 AM — quietSep 19, 5 AM — quietSep 19, 8 AM — quietSep 19, 11 AM — quietSep 19, 2 PM — quietSep 19, 5 PM — quietSep 19, 8 PM — quietSep 19, 11 PM — quietSep 20, 2 AM — quietSep 20, 5 AM — quietSep 20, 8 AM — quietSep 20, 11 AM — quietSep 20, 2 PM — quietSep 20, 5 PM — quietSep 20, 8 PM — quietSep 20, 11 PM — quietSep 21, 2 AM — quietSep 21, 5 AM — quietSep 21, 8 AM — quietSep 21, 11 AM — quietSep 21, 2 PM — quietSep 21, 5 PM — quietSep 21, 8 PM — 1 piece · 1 post — Hacker News 1Sep 21, 11 PM — quietSep 22, 2 AM — quietSep 22, 5 AM — 1 piece · 1 post — Hacker News 1Sep 22, 8 AM — quietSep 22, 11 AM — quietSep 22, 2 PM — quietSep 22, 5 PM — quietSep 22, 8 PM — quietSep 22, 11 PM — 2 pieces · 2 articles — Newswires 2Sep 23, 2 AM — 1 piece · 1 post — Hacker News 1Sep 23, 5 AM — quietSep 23, 8 AM — 1 piece · 1 post — Hacker News 1Sep 23, 11 AM — 2 pieces · 2 posts — Hacker News 2Sep 23, 2 PM — quietSep 23, 5 PM — quietSep 23, 8 PM — quietSep 23, 11 PM — quietSep 24, 2 AM — quietSep 24, 5 AM — 2 pieces · 2 posts — Hacker News 2Sep 24, 8 AM — quietSep 24, 11 AM — quietSep 24, 2 PM — quietSep 24, 5 PM — quietSep 24, 8 PM — quietSep 24, 11 PM — quietSep 25, 2 AM — quietSep 25, 5 AM — quietSep 25, 8 AM — quietSep 25, 11 AM — quietSep 25, 2 PM — quietSep 25, 5 PM — quietSep 25, 8 PM — quietSep 25, 11 PM — quietYesterday, 2 AM — quietYesterday, 5 AM — quietYesterday, 8 AM — quietYesterday, 11 AM — quietYesterday, 2 PM — quietYesterday, 5 PM — quietYesterday, 8 PM — quietYesterday, 11 PM — quietToday, 2 AM — quietToday, 5 AM — quietToday, 8 AM — quietToday, 11 AM — quietToday, 2 PM — quiet 123
Sep 18Sep 19Sep 20Sep 21Sep 22Sep 23Sep 24Sep 25yesterdaynow · 3:49 PM ET
  1. 3

    SoL-Pi findings show 44–49% token reduction and ~$5–$13 hourly savings

    Detailed results from the SoL-Pi study show token traffic falls 44.7–49.0%, API costs drop by roughly one third, with estimated hourly savings of $8.75–$13.50 against native Codex and Claude Code harnesses and $4.36–$5.71 against the Pi baseline. Performance matches across GPT-5.6 Sol and Opus 5 on the 51-task EdgeBench evaluation.

    “estimated hourly savings are $8.75-$13.50 relative to native Codex and Claude Code harnesses, and $4.36-$5.71 relative to Pi.”
    — SoL-Pi researchers, Authors · source
    • Something that totally confuses me is the use of the term “harness.”Is it the harness that enables the LLM to use tools and implement plans, or is it the short leash that, through many guardrails, ensures that the LLM follows the desired path?I see both meanings used. The former is still plausible, but I see many people using the second…

      kaufmannHacker News8d agoview on Hacker News ↗
    2 more of the top 3 · 4 posts in this stretch
    • Planning and tooling are suitably designed for execution. That's why they fail to decide on the resulting coding accuracy. If the product is user-facing (i.e. all products), a blind verification harness that is specifically designed to behave like a real user should be the decider.

      hosamshHacker News8d agoview on Hacker News ↗
    • Love your analogy.Cars performance is hugely dependent on use case and overall setup (suspension, engine (NA, turbo, supercharger, etc), coef of drag, etc).Maps well onto the LLM surface.

      BIGFOOT_EXISTSHacker News9d agoview on Hacker News ↗
    all of them →
  2. 2

    Research community discusses harness design and benchmarking trade-offs

    Hacker News users engage with the empirical harness study, debating the generalizability of findings across model families, the definition of terms like 'bash capable' and 'harness,' and whether existing benchmarks capture real-world performance differences.

    “we definitely need more principled studies on the role of harnesses. I'd also say that there aren't too many benchmarks where the more complicated harnesses consistently outperform extremely simple agents.”
    — lieret
    • The conclusions:> Planning improves success at additional cost for weaker models but mainly reduces cost, with small decreases in success rate, for stronger models.> Predefined tools raise success rates for models with weak bash control, whereas bash-only yields higher success at lower cost for bash-capable models, most clearly on shell-centric…

      embedding-shapeHacker News9d agoview on Hacker News ↗
    2 more of the top 3 · 8 posts in this stretch
    • Haven't gone through full PDF as its very detailed, few things have resonated with me so far.Basically if a Car A is performing better (be it speed, milage or in general sense) than Car B, then it is not necessarily because its engine. It could be because of better tires, better gearbox, lighter body, better usability of features, etc.You can…

      gps372Hacker News9d agoview on Hacker News ↗
    • Quite inline with what I had found with my Claude code sessions over the last year. I wrote about this a few months ago.https://rahulmax.com/notes/how-i-keep-the-ai-bill-down/In their case, context management pays off more the tighter your window. Their gap between managing and not managing is 35.7 points of success rate at 32k and 2.7 points at…

      rahulmaxHacker News9d agoview on Hacker News ↗
    all of them →
  3. 1

    Red-teaming study examines blocking classifiers for malign coding agents

    Alex Remedios and colleagues release research on red-teaming auto mode—the production blocking monitors in Claude Code and OpenAI's Codex—testing their robustness when agents are persistently misaligned, not just accidentally harmful.

    1. first by HN Best, 9d ago · also HN Frontpage, arXiv cs.AI

    2. 1 outlet first by arXiv cs.AI, 9d ago · read ↗

  4. background

    Empirical study of harness components appears on arXiv — Run-Ze Fan and colleagues publish an empirical study examining individual components of coding harnesses—planning, action space, and other elements—to understand which factors drive coding agent performance, moving beyond treating harnesses as monolithic systems.

  5. background

    Researchers publish SoL-Pi harness optimization study — Haozhe Liu and colleagues at NVIDIA, NTU, and MIT announce SoL-Pi, which applies recursive auto-research loops at the harness layer to optimize coding agent efficiency. The work identifies four mechanisms—Action Fusion, Online Context Compact, ObservationPack, and Evidence-Preserving Reducer—that survive selection and transfer across diverse environments.

What people are saying 6 voices from 1 site · best of 12 · verbatim

Still unanswered
  • What does 'bash capable' mean formally, and how do you measure it?
  • Do SoL-Pi's findings transfer to open-source models and to the current frontier (Qwen, DeepSeek, latest Claude/GPT)?
  • How much of the measured improvement actually matters in user-facing products, versus lab benchmarks?