Skip to content
VECTOR WIREAI INTELLIGENCE
⌘K
UTC
Models741Deals30d262Policy30d64, 11 pendingSources1,015

Terminal-Bench Hard leaderboard — the family board.

7 TRACKED VARIANTS · 5 FIRST-CLASS BOARDS

One family, 7 tracked prints: Terminal-Bench 2.1, Terminal-Bench 2.0, Terminal-Bench and Terminal-bench 3.0 are cited standalone and keep live boards of their own. Scoring configs and splits of the base board render as sections of this page, each under its own honest label.

● LIVE DATA288 MODELS ON THIS BOARDLEADER GPT-5.3 CODEX AT 77.3

First-class boards in this familyCITED STANDALONE · EACH KEEPS ITS OWN LIVE BOARD

#BoardModels scored
01Terminal-Bench HardTHIS PAGE288
02Terminal-Bench 2.1184
03Terminal-Bench 2.066
04Terminal-Bench25
05Terminal-bench 3.014

The board — Terminal-Bench HardTOP 100 OF 288 MEASURED · BEST SCORE PER MODEL, DEFAULT VARIANT

#ModelVendorScoreVLast seen
01GPT-5.3 CodexOpenAI77.32026-05-02
02Gemini 3.1 ProGoogle68.52026-05-02
03GPT-5.6 SolOpenAI65.92026-07-10
04Claude Opus 4.6Anthropic65.42026-10-04
05GPT-5.6 TerraOpenAI62.92026-07-10
06Claude Fable 5Anthropic62.92026-10-04
07GPT-5.2OpenAI62.22026-05-02
08GPT-5.5OpenAI60.62026-06-01
09Claude Sonnet 4.6Anthropic59.12026-10-04
10Claude Opus 4.8Anthropic58.32026-10-04
11GPT-5.4OpenAI57.62026-05-02
12Gemini 3 ProGoogle56.92026-06-13
13Claude Opus 4.7Anthropic54.52026-10-04
14GPT-5.4 miniOpenAI52.32026-06-01
15Qwen3.7 MaxAlibaba50.82026-05-20
16GLM-5.2Z.ai50.82026-06-17
17Qwen3.7 Plus PreviewAlibaba47.02026-06-04
18Claude Opus 4.5Anthropic47.02026-05-06
19DeepSeek-V4-ProDeepSeek46.22026-10-01
20Gemini 3.5 FlashGoogle46.22026-09-10
21GPT-5.1OpenAI45.52026-06-20
22Muse SparkMeta45.52026-05-02
23Kimi K2.7 CodeMoonshot44.72026-06-17
24Qwen3.6 PlusAlibaba43.92026-05-02
25Kimi K2.6Moonshot43.92026-05-02
26Qwen3.6 Max PreviewAlibaba43.92026-05-02
27GLM-5Z.ai43.22026-05-02
28GLM-5.1Z.ai43.22026-05-02
29MiMo V2.5 ProXiaomi43.22026-09-22
30MiniMax M3MiniMax42.42026-06-04
31GPT-5.4 nanoOpenAI42.42026-09-10
32GPT-5.5 InstantOpenAI42.42026-05-31
33MiMo V2.5Xiaomi41.72026-09-24
34Qwen3.5 397B A17BAlibaba40.92026-06-01
35Grok 4.20SpaceXAI40.92026-09-10
36MiMo V2 ProXiaomi40.92026-05-02
37MiniMax M2.7MiniMax39.42026-05-02
38Gemini 3 Flash PreviewGoogle38.62026-05-02
39DeepSeek-V4-FlashDeepSeek38.62026-10-01
40GPT-5OpenAI37.92026-06-21
41GPT-5 CodexOpenAI37.92026-05-02
42Grok 4SpaceXAI37.92026-05-02
43Grok 4.3SpaceXAI37.92026-05-23
44GPT-5.2 CodexOpenAI37.12026-05-02
45o3OpenAI37.12026-05-02
46Nemotron 3 Ultra 550B A55BNVIDIA36.42026-06-06
47Gemma 4 31BGoogle36.42026-06-01
48Claude Sonnet 4.5Anthropic35.62026-06-06
49MiMo V2 OmniXiaomi35.62026-05-02
50DeepSeek-V3.2DeepSeek35.62026-06-13
51Step 3.7 FlashStepFun35.62026-06-02
52Qwen3.6 35B A3BAlibaba34.82026-06-01
53MiniMax M2.5MiniMax34.82026-05-02
54DeepSeek-V3.2-SpecialeDeepSeek34.82026-05-02
55GPT-5.1 CodexOpenAI34.82026-05-02
56Qwen3.6 27BAlibaba34.82026-05-02
57Kimi K2.5Moonshot34.82026-06-01
58Claude Opus 4.1Anthropic34.32026-06-01
59GPT-5 miniOpenAI33.32026-06-21
60GLM-5-TurboZ.ai33.32026-05-02
61GPT-5.1 Codex miniOpenAI33.32026-05-02
62Mistral Medium 3.5 128BMistral33.32026-05-02
63GLM-4.7Z.ai33.32026-06-01
64Qwen3.5 27BAlibaba32.62026-05-02
65Step 3.5 FlashStepFun32.62026-06-18
66GLM 5V TurboZ.ai32.62026-06-01
67DeepSeek-V3.1-TerminusDeepSeek31.82026-06-01
68Hy3-previewTencent31.82026-05-02
69North-Mini-Code-1.0Cohere31.12026-08-23
70DeepSeek-V3.2-ExpDeepSeek31.12026-06-01
71Qwen3.5 122B A10BAlibaba31.12026-05-02
72Kimi K2 ThinkingMoonshot31.12026-06-13
73Claude Opus 4Anthropic31.12026-06-01
74ling-2-6-1tInclusionAI31.12026-05-02
75MiMo V2 FlashXiaomi31.12026-06-13
76Claude Sonnet 4Anthropic31.12026-06-23
77JT-35B-FlashChina Mobile28.82026-05-13
78GLM-4.6Z.ai28.82026-06-19
79MiniMax M2.1MiniMax28.82026-05-02
80Nemotron 3 Super 120B A12BNVIDIA28.82026-09-10
81Claude Haiku 4.5Anthropic27.32026-07-20
82Doubao Seed CodeByteDance26.52026-05-02
83Mercury 2Inception26.52026-05-02
84Gemini 2.5 ProGoogle26.52026-06-06
85Qwen3.5 35B A3BAlibaba26.52026-05-02
86MiniMax M2MiniMax25.82026-06-06
87DeepSeek-V3.1DeepSeek252026-06-01
88Command A+Cohere252026-08-23
89Gemma 4 26B A4BGoogle252026-05-06
90ERNIE 5.0 Thinking PreviewBaidu252026-05-02
91Qwen3.5 9BAlibaba24.22026-05-02
92Grok 4.1 FastSpaceXAI24.22026-06-23
93Nova 2.0 Pro PreviewAmazon24.22026-05-02
94Qwen3 Max (Reasoning)Alibaba24.22026-09-22
95Gemini 3.1 Flash Lite PreviewGoogle24.22026-05-02
96gpt-oss-120bOpenAI23.52026-05-02
97Kimi K2 InstructMoonshot23.52026-06-06
98K-EXAONELG AI22.72026-05-02
99Trinity Large ThinkingArcee AI22.72026-05-02
100GLM-4.5Z.ai22.02026-06-01
◆ IV — INDEPENDENTLY VERIFIED · ◆ AA — AGGREGATOR · ◆ VA — VENDOR · ◇ SA — SOURCE-ATTRIBUTED

Variants of this boardSCORING CONFIGS & SPLITS — NOT CITED STANDALONE, SO THEY LIVE HERE

Terminal-Bench w/ simulated tools (JSON)

VARIANT5 models · best score per model
#ModelVendorScoreVLast seen
01Claude Sonnet 4.5Anthropic512026-06-12
02Kimi K2 ThinkingMoonshot47.12026-06-12
03Kimi K2 InstructMoonshot44.52026-06-12
04GPT-5OpenAI43.82026-06-12
05DeepSeek-V3.2DeepSeek37.72026-06-12

Other tracked prints

THIN BOARDSunder 5 published models each — listed, never hidden
PrintLeaderScoreModels
Terminal Bench (hard subset)Nemotron 3 Super 120B A12B25.52
THE FAMILY BOARD · EVERY PRINT LABELED · EVERY SCORE SOURCE-ATTRIBUTED ON ITS OWN LIVE BOARDAll leaderboardsThe frontier boardModels by vendor