← The Wire

Public story · 2026-02-24 · source-backed

SWE-bench February Analysis: The 80% Ceiling and What It Means

Confidence
source-backed
Sources
1
Redaction
passed

Story

simonwillison.net — Four models now above 80% SWE-bench (Opus 4.6 at 80.8%, Gemini 3.1 Pro at 80.6%, GPT-5.3-Codex at 80.0%, Kimi K2.5 at 76.8%). Willison's analysis: we're approaching the ceiling where SWE-bench stops being a meaningful differentiator. The gap between models is now smaller than the gap between scaffolding approaches. This is the quantitative backing for the "scaffolding > model" thesis.

Related stories

Each link below shares sources, entities, or timing with this story.

  1. Shared entities / Shared topic

    LogRocket February Power Rankings: Open-Source Breaks Into Top 5

    Both cover Codex, Gemini, GPT, Kimi K2; overlapping topics (kimi, model, swe-bench).

  2. Shared entities / Same source domain / Shared topic / What happened next / Tension

    GPT-5.5 Is Here. It's Fast, It's Expensive, and It Changes How You Prompt.

    Both cover Codex, GPT, Opus, SWE; reported by the same outlet (simonwillison.net); overlapping topics (gpt-5, model).

  3. GPT-5.4 Mini ($0.75/M) and Nano ($0.20/M) — Purpose-Built for Subagent Workloads

    Both cover Gemini, GPT, Opus, SWE; reported by the same outlet (simonwillison.net); overlapping topics (gpt-5, model).

  4. Shared entities / Shared topic / What happened next / Tension

    Kimi K2.6: A 1-Trillion-Parameter Open-Weights Model Just Beat GPT-5.4 on SWE-Bench Pro

    Both cover GPT, Kimi K2, Opus, SWE; overlapping topics (gpt-5, model, swe-bench); picks up the GPT thread on 2026-04-21.

  5. Shared entities / Shared topic / What happened next

    June 2026 was the biggest model launch month ever, and ChatGPT just dropped below 50% share

    Both cover Gemini, GPT, Kimi K2, Opus; overlapping topics (gemini, gpt-5, model); picks up the Gemini thread on 2026-06-21.

  6. Shared entities / Shared topic / What happened next / Tension

    Kimi K2.6 Beats GPT-5.4 on SWE-Bench Pro. It's Open-Weight. It Costs 88% Less.

    Both cover GPT, Kimi K2, SWE; overlapping topics (gpt-5, kimi, model, swe-bench); picks up the GPT thread on 2026-05-05.

  7. Shared entities / Same source domain / Shared topic / What happened next

    GLM-5.1: Open-Weight 754B Model Takes #1 on SWE-Bench Pro, Runs 8 Hours Unsupervised. MIT Licensed.

    Both cover Gemini, GPT, SWE; reported by the same outlet (simonwillison.net); overlapping topics (above, model, swe-bench).

  8. Shared entities / Shared topic / What happened next

    Open-weight frontier coding models are now benchmark-competitive with closed.

    Both cover Gemini, GPT, Kimi K2, SWE; overlapping topics (gemini, model); picks up the Gemini thread on 2026-06-08.

Source trail

Entities

Provenance

AI generated
no
Story unit
2026-02-24-swe-bench-february-analysis-the-80-ceiling-and-what-it-means
Labels
source-backed, canonical briefing excerpt
SWE-bench February Analysis: The 80% Ceiling and What It Means | MindPattern