← The Wire

Vibe Coding2026-06-08 · source-backed

Agentic leaderboards are shifting from raw SWE-Bench to blended tool-use benchmarks.

Confidence
source-backed
Sources
1
Redaction
redacted

Story

June leaderboards increasingly score a weighted blend of Terminal-Bench 2.0, BrowseComp, and OSWorld-Verified instead of SWE-bench alone, with BenchLM weighting "agentic" browse-and-do workflows at 22%, its highest single category. When you evaluate a tool, the headline chat or single-file coding score understates real agent performance. What matters now is how the model executes multi-step, tool-using, environment-acting work end to end. Stop reading the one number on the model card.

Related stories

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

  1. Shared entities / Shared topic / Earlier coverage

    Berkeley Researchers Score 100% on SWE-bench Verified Without Solving a Single Task. Every Major AI Benchmark Is Broken.

    Both cover Bench, OSWorld, SWE, Terminal; overlapping topics (agent, benchmark, coding, model, score); earlier Bench coverage from 2026-04-12.

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

    Cohere open-sourced a 30B coding model that runs on a single H100. Agentic coding without the frontier bill.

    Both cover Agentic, Bench, SWE, Terminal; overlapping topics (agent, agentic, coding, model); picks up the Agentic thread on 2026-06-10.

  3. Shared entities / Shared topic / What happened next

    Kimi K2.7-Code: a 1T-param open coding model with benchmarks nobody can check

    Both cover Bench, SWE, Terminal, Verified; overlapping topics (agentic, benchmark, coding, model, swe-bench); picks up the Bench thread on 2026-06-14.

  4. Shared entities / Shared topic / Earlier coverage / Tension

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

    Both cover Bench, SWE, Terminal; overlapping topics (agent, agentic, benchmark, coding, model); earlier Bench coverage from 2026-04-21.

  5. Shared entities / Shared topic / What happened next

    GLM-5.2 is eating OpenRouter, and it's filling the hole Fable 5 left

    Both cover Bench, SWE, Terminal, When; overlapping topics (agentic, coding, model, swe-bench); picks up the Bench thread on 2026-06-25.

  6. Shared entities / Shared topic / Earlier coverage

    Poolside Ships a 68.2% SWE-bench Coding Agent That Runs on Your Mac. Apache 2.0.

    Both cover Bench, SWE, Terminal, Verified; overlapping topics (agent, coding, model, swe-bench); earlier Bench coverage from 2026-04-29.

  7. GPT-5.4 Launches: Computer Use, Tool Search API, 1M Context

    Both cover BrowseComp, OSWorld, SWE, Verified; overlapping topics (agent, benchmark, browsecomp, model); earlier BrowseComp coverage from 2026-03-06.

  8. GLM-5: 744B Open-Source, Free on NVIDIA NIM

    Both cover Bench, SWE, Terminal, Verified; overlapping topics (benchmark, card, coding, model); earlier Bench coverage from 2026-03-04.

Source trail

Entities

Provenance

AI generated
no
Story unit
2026-06-08-agentic-leaderboards-are-shifting-from-raw-swe-bench-to-blended-tool-use-benchmarks
Labels
source-backed, canonical briefing excerpt