Skip to content
Yardstick
  1. Benchmarks
  2. Terminal-Bench 4.0

Terminal-Bench 4.0

Realistic command-line jobs in a Linux container — build, debug, analyse, administer — each checked by tests.

CodingRun by Terminal-Bench teamOfficial leaderboard (opens in a new tab)

Top score
64.8%Claude Opus 5.5
Median of field
39.2%13 models scored
Data as of
Oct 6, 2026tbench.ai leaderboard 4.0

Standings

Bars share one axis from 0% to 100%.

  1. Rank 1
    Claude Opus 5.564.8%

    Anthropic · $4 / $20New

  2. Rank 2
    Claude Sonnet 5.561.8%

    Anthropic · $2 / $10New

  3. Rank 3
    GPT-6 Astra58.2%

    OpenAI · $10 / $50New

  4. Rank 3
    GPT-6.1 Sol58.2%

    OpenAI · $2 / $10New

  5. Rank 5
    Claude Fable 5.157.9%

    Anthropic · $10 / $50New

  6. Rank 6
    GLM-5.341.8%

    Z.ai · $1.40 / $4.40NewOpen weights

  7. Rank 7
    Claude Haiku 5.539.2%† (reported by the lab)

    Anthropic · $0.10 / $0.50New

  8. Rank 8
    Grok 4.737.6%

    SpaceXAI · $2 / $6New

  9. Rank 9
    GLM-5.3-Flash35.8%

    Z.ai · $0.15 / $0.50NewOpen weights

  10. Rank 10
    Qwen3.8-Max27%

    Alibaba · $2 / $6

  11. Rank 11
    Gemini 3.8 Flash19.1%

    Google · $0.75 / $3.75New

  12. Rank 12
    GPT-6 Luna16.4%

    OpenAI · $0.10 / $0.50New

  13. Rank 13
    Muse Spark 1.314.6%

    Meta · $1.25 / $4.25New

Not reported for Terminal-Bench 4.0 · 16

  • ·
    DeepSeek-V4-Pro—Not reported

    DeepSeek · $1.32 / $3.96NewOpen weights

  • ·
    DeepSeek-V4.1-Flash—Not reported

    DeepSeek · $0.30 / $1.20NewOpen weights

  • ·
    Gemini 3.1 Pro—Not reported

    Google · $2 / $12Preview

  • ·
    Gemini 3.5 Flash-Lite—Not reported

    Google · $0.30 / $2.50

  • ·
    Gemini 4 Argon—Not reported

    Google · $2 / $10NewPreview

  • ·
    Gemma 4 31B—Not reported

    Google · Self-hostedOpen weights

  • ·
    Kimi K2.7 Code—Not reported

    Moonshot AI · $0.95 / $4Open weights

  • ·
    Kimi K3—Not reported

    Moonshot AI · $3 / $15Open weights

  • ·
    MiMo-V2.6-Pro—Not reported

    Xiaomi · $0.435 / $0.87NewOpen weights

  • ·
    MiniMax M3—Not reported

    MiniMax · $0.30 / $1.20Open weights

  • ·
    Mistral Large 4—Not reported

    Mistral AI · $1.36 / $4.18NewPreview

  • ·
    Mistral Medium 3.5—Not reported

    Mistral AI · $1.50 / $7.50Open weights

  • ·
    Mistral Small 4—Not reported

    Mistral AI · $0.15 / $0.60Open weights

  • ·
    Muse Glimmer 30B—Not reported

    Meta · Self-hostedNewOpen weights

  • ·
    Qwen3.8-27B—Not reported

    Alibaba · $0.50 / $3NewOpen weights

  • ·
    Qwen3.8-Flash-Next—Not reported

    Alibaba · $0.15 / $0.47NewOpen weights

† Reported by the model’s own lab and not independently reproduced. Unmarked scores come from the benchmark’s maintainer or a third-party evaluator.

What it measures

Whether a coding agent can finish realistic terminal work end to end: building and debugging software, data and machine-learning workflows, and system and security tasks.

How it is scored

Percent of trials solved, each in a fresh container with an eight-hour limit. Every row pairs a model with an agent harness, usually the lab's own (Claude Code, Codex, Grok Build); this site lists each model's best row.

Read it with care

The score reflects model and harness together, and labs run different harnesses. Version 4.0 (28 August 2026) removed tasks the maintainers considered saturated, so it cannot be compared with 2.x results.

Other benchmarks

  • Artificial Analysis Intelligence Index
  • Epoch Capabilities Index
  • LiveBench
  • SWE-Bench Pro (V2 Hard)
  • Arena WebDev
  • GDPval-AA
  • Vending-Bench 2
  • ARC-AGI-2
  • ARC-AGI-3
  • FrontierMath (Tiers 1–3)
  • MathArena
  • SimpleBench
  • Humanity's Last Exam (Diamond)
  • GPQA Diamond
  • Arena Vision
  • Arena Text
Yardstick

A plain reading of how today’s SI models compare. Scores are taken from each benchmark’s own leaderboard wherever one exists; anything a lab reported about its own model is marked.

Data as of Oct 8, 2026.

Explore

  • Models
  • Compare
  • Benchmarks
  • Value
  • Labs
  • Methodology & sources

Sources

  • Artificial Analysis (opens in a new tab)
  • Arena (opens in a new tab)
  • Epoch AI Benchmarking Hub (opens in a new tab)
  • ARC Prize (opens in a new tab)
  • Terminal-Bench (opens in a new tab)
  • Scale AI Labs (opens in a new tab)

Yardstick is an independent reference and is not affiliated with, sponsored by or endorsed by any lab listed. Product names and logos are trademarks of their respective owners and are used only to identify them. Benchmarks are imperfect proxies: check a model on your own task before you commit to it.

  • Models
  • Compare
  • Benchmarks
  • Value
  • Labs
  • Models
  • Compare
  • Benchmarks
  • Value
  • Labs
  • Models
  • Compare
  • Benchmarks
  • Value
  • Labs