BenchLeader
Menu
Best LLMsChoose a modelCoding plansLeaderboardsModelsBenchmarksProvidersReleases7

Terminal-Bench 4.0 (Vals) leaderboard

As of 6 Nov 2026, the top score on Terminal-Bench 4.0 (Vals) is 65.2% by Claude Opus 5.5 (Anthropic), ahead of Claude Sonnet 5.5 at 64.1%, across 59 model configurations with a published result (Vals AI, result dated 5 Nov 2026). Terminal-Bench 4.0 tasks written to defeat current frontier agents, run by Vals AI. Terminal-Bench 4.0 (Vals) is shown for reference only and does not count toward the BenchLeader Index.

Key facts (as of 6 Nov 2026)

  • Claude Opus 5.5 holds the top Terminal-Bench 4.0 (Vals) score at 65.2% (Vals AI, 5 Nov 2026).
  • Claude Sonnet 5.5 is second on Terminal-Bench 4.0 (Vals) at 64.1% (Vals AI, 5 Nov 2026).
  • Terminal-Bench 4.0 (Vals): Terminal-Bench 4.0, the frontier-difficulty set, in Vals AI's harness. Run by Vals AI (Vals AI).
  • 59 model configurations have a published Terminal-Bench 4.0 (Vals) result on BenchLeader (BenchLeader, 6 Nov 2026).
  • Terminal-Bench 4.0 (Vals) does not feed the BenchLeader Index; it is shown for reference (BenchLeader methodology).

Terminal-Bench 4.0, the frontier-difficulty set, in Vals AI's harness. Run by Vals AI.

Published by
Vals AI
Category
Agents & tools
Index weight
Reference only
Models
59
Data as of
6 Nov 2026

Vals AI (vals.ai).

What the test looks like

Terminal-Bench 4.0 tasks written to defeat current frontier agents, run by Vals AI.

How it is scored

Percent of tasks passing, run by Vals AI.

What to keep in mind

New and hard; small set, low scores.

Top 15
  1. 1Claude Opus 5.565.2%
  2. 2Claude Sonnet 5.564.1%
  3. 3Kimi K3.1 Pro (max)63.6%
  4. 4Gemini 4 Carbon (high)63.1%
  5. 5GPT-6 Astra (max)59.6%
  6. 6Claude Fable 5.158.1%
  7. 7Muse Wonder Full (max)58.1%
  8. 8Gemini 4 Argon (high)57.6%
  9. 9GPT-6.1 Sol (max)55.0%
  10. 10Claude Opus 553.5%
  11. 11GPT-6.1 Terra (max)49.5%
  12. 12Muse Wonder (max)49.5%
  13. 13GPT-6 Sol (max)44.4%
  14. 14Claude Fable 541.4%
  15. 15Hy439.4%
59 of 59
#
1Claude Opus 5.565.2%63.4$13.202026-11-05
2Claude Sonnet 5.564.1%–$16.512026-11-05
3Kimi K3.1 Promax63.6%–$4.572026-11-05
4Gemini 4 Carbonhigh63.1%69.2$7.452026-11-05
5GPT-6 Astramax59.6%67.8$9.582026-11-05
6Claude Fable 5.158.1%66.4$17.182026-11-05
7Muse Wonder Fullmax58.1%62.9$10.532026-11-05
8Gemini 4 Argonhigh57.6%66.0$17.642026-11-05
9GPT-6.1 Solmax55.0%67.3$1.722026-11-05
10Claude Opus 553.5%65.2$18.602026-11-05
11GPT-6.1 Terramax49.5%65.3$2.162026-11-05
12Muse Wondermax49.5%62.3$14.662026-11-05
13GPT-6 Solmax44.4%63.3$5.792026-11-05
14Claude Fable 541.4%66.5$30.342026-11-05
15Hy439.4%55.7$1.282026-11-05
16GLM 5.3max38.9%59.0$9.372026-11-05
17Qwen3.8 Max Internal38.9%–$12.152026-11-05
18GPT-5.6 Solmax37.9%65.4$7.982026-11-05
19GLM 5.3 Flashmax37.4%54.2$1.222026-11-05
20Claude Haiku 5.536.4%–$3.122026-11-05
21Qwen3.8 Maxmax34.3%58.8$10.632026-11-05
22DeepSeek V4.1 Flash 1017max31.8%59.2$0.5392026-11-05
23Step 5 Preview31.8%54.9$3.112026-11-05
24MiMo-V2.6-Pro31.3%57.3$0.4572026-11-05
25Grok 4.7xhigh28.8%58.3$18.012026-11-05
26Mistral Large 4high27.8%49.5$5.182026-11-05
27Mistral Large 4 OSShigh26.3%–$2.732026-11-05
28MiniMax M3.1 Flash25.3%53.0$0.7352026-11-05
29Muse Spark 1.3max24.8%62.0$6.652026-11-05
30MiMo-V2.6-Flash24.2%55.0$0.2122026-11-05
31Claude Opus 4.823.2%60.0$17.142026-11-05
32GPT-5.6 Terramax22.7%60.7$5.602026-11-05
33Ember-120.7%–$19.482026-11-05
34Gemini 3.8 Flashhigh19.7%62.0$8.652026-11-05
35DeepSeek V4.1 Flashhigh19.7%55.7$0.4992026-11-05
36DeepSeek V4 Flashhigh18.7%57.3$0.7662026-11-05
37Kimi K3max17.2%62.1$12.482026-11-05
38Grok 4.6high17.2%60.5$5.062026-11-05
39DeepSeek V4 Promax14.1%57.0$3.312026-11-05
40GPT-6 Lunamax13.6%56.7$0.3552026-11-05
41NewInkling 2 Small12.6%57.4$8.222026-11-05
42Gemini 3.7 Flashhigh12.1%62.2$8.712026-11-05
43GPT-5.6 Lunamax11.6%56.3$0.7322026-11-05
44Muse Spark 1.3xhigh10.6%58.5$7.042026-11-05
45Claude Sonnet 510.6%55.9$26.052026-11-05
46Grok 4.5high8.6%56.9$5.522026-11-05
47Muse Spark Openxhigh6.6%54.0$7.262026-11-05
48Gemini 3.5 Flashhigh6.1%62.5$5.982026-11-05
49Muse Spark 1.2xhigh6.1%61.7$4.142026-11-05
50Beam 1.04.0%45.8–2026-11-05
51Mercury 3high3.0%–$0.4222026-11-05
52Gemini 3.1 Prohigh2.5%60.1$4.112026-11-05
53GPT-5.4 Minixhigh2.5%56.0$1.442026-11-05
54MiniMax M31.0%56.4$7.232026-11-05
55MiMo-V2.5-Pro0.5%54.9$0.2332026-11-05
56Inkling0.5%47.5$2.502026-11-05
57Inkling Small0.0%52.1$0.3222026-11-05
58Mercury 2.5high0.0%–$0.3682026-11-05
59Nemotron 3.5 Super VL 120B A12Bhigh0.0%–$0.8632026-11-05

More agents & tools benchmarks · All LLM benchmarks

Related benchmarks

Top head-to-heads on Terminal-Bench 4.0 (Vals)

Cite as: BenchLeader, “Terminal-Bench 4.0 (Vals) leaderboard”, data as of 6 Nov 2026, https://www.benchleader.com/benchmarks/vals_terminal_bench_4.

Terminal-Bench 4.0 (Vals): questions

What does Terminal-Bench 4.0 (Vals) measure?
Terminal-Bench 4.0 tasks written to defeat current frontier agents, run by Vals AI. Scores are reported in percent of tasks solved; higher is better.
What is the top score on Terminal-Bench 4.0 (Vals)?
As of 6 Nov 2026, the highest score on Terminal-Bench 4.0 (Vals) is 65.2% by Claude Opus 5.5 (result dated 5 Nov 2026), ahead of Claude Sonnet 5.5 at 64.1%. Scores come from Vals AI and are re-read daily.
Which AI model leads Terminal-Bench 4.0 (Vals)?
Claude Opus 5.5 leads Terminal-Bench 4.0 (Vals) with 65.2% as of 6 Nov 2026 (result dated 5 Nov 2026), ahead of Claude Sonnet 5.5 at 64.1%.
How many models have Terminal-Bench 4.0 (Vals) results?
59 model configurations have a Terminal-Bench 4.0 (Vals) result on BenchLeader, all taken from Vals AI.
Who runs Terminal-Bench 4.0 (Vals) and how often is it updated?
Terminal-Bench 4.0 (Vals) is published by Vals AI. BenchLeader re-reads the published results once a day and records the date each result was published.
Does Terminal-Bench 4.0 (Vals) count toward the BenchLeader Index?
No. Terminal-Bench 4.0 (Vals) is shown for reference but left out of the composite index.