Skip to content

Benchmark

Workspace-Bench-Lite · Total

Scores are percentages. Higher is better.

Models and systems measured
10
Scores recorded
45
Sources
1
IQ weight
0.02

Score distribution

45.5%
Mean
45.9%
Median
10.4%
Std dev
34.4%
10th percentile
57.7%
90th percentile
13.8%
Min
13.8%66.7%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 0.02 for chat models, measured over 7 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#Model or systemScore
1Claude Opus 4.7OpenClaw · Workspace-Bench-Lite · 100 tasks · rubric pass rate· Reported by Workspace-Bench· via BenchBoard docs66.7%
2Claude Opus 4.7ClaudeCode · Workspace-Bench-Lite · 100 tasks · rubric pass rate· Reported by Workspace-Bench· via BenchBoard docs64.7%
3Claude Opus 4.7Hermes · Workspace-Bench-Lite · 100 tasks · rubric pass rate· Reported by Workspace-Bench· via BenchBoard docs64.5%
4GLM 5.1DeepAgent · Workspace-Bench-Lite · 100 tasks · rubric pass rate· Reported by Workspace-Bench· via BenchBoard docs61%
5GLM 5.1Hermes · Workspace-Bench-Lite · 100 tasks · rubric pass rate· Reported by Workspace-Bench· via BenchBoard docs57.7%
6GLM 5.1OpenClaw · Workspace-Bench-Lite · 100 tasks · rubric pass rate· Reported by Workspace-Bench· via BenchBoard docs57.5%
7Qwen3.6 PlusOpenClaw · Workspace-Bench-Lite · 100 tasks · rubric pass rate· Reported by Workspace-Bench· via BenchBoard docs55.6%
8MiniMax M2.7ClaudeCode · Workspace-Bench-Lite · 100 tasks · rubric pass rate· Reported by Workspace-Bench· via BenchBoard docs54.6%
9Claude Opus 4.7DeepAgent · Workspace-Bench-Lite · 100 tasks · rubric pass rate· Reported by Workspace-Bench· via BenchBoard docs54.4%
10GLM 5.1Codex · Workspace-Bench-Lite · 100 tasks · rubric pass rate· Reported by Workspace-Bench· via BenchBoard docs53.1%
11GLM 5.1ClaudeCode · Workspace-Bench-Lite · 100 tasks · rubric pass rate· Reported by Workspace-Bench· via BenchBoard docs52.6%
12MiniMax M2.7Hermes · Workspace-Bench-Lite · 100 tasks · rubric pass rate· Reported by Workspace-Bench· via BenchBoard docs52.6%
13GPT 5.4ClaudeCode · Workspace-Bench-Lite · 100 tasks · rubric pass rate· Reported by Workspace-Bench· via BenchBoard docs51.8%
14Qwen3.6 PlusHermes · Workspace-Bench-Lite · 100 tasks · rubric pass rate· Reported by Workspace-Bench· via BenchBoard docs50.9%
15Kimi K2.5Hermes · Workspace-Bench-Lite · 100 tasks · rubric pass rate· Reported by Workspace-Bench· via BenchBoard docs49.1%
16Kimi K2.5ClaudeCode · Workspace-Bench-Lite · 100 tasks · rubric pass rate· Reported by Workspace-Bench· via BenchBoard docs48.3%
17GPT 5.4Codex · Workspace-Bench-Lite · 100 tasks · rubric pass rate· Reported by Workspace-Bench· via BenchBoard docs47.7%
18Qwen3.6 PlusCodex · Workspace-Bench-Lite · 100 tasks · rubric pass rate· Reported by Workspace-Bench· via BenchBoard docs47.3%
19GPT 5.4OpenClaw · Workspace-Bench-Lite · 100 tasks · rubric pass rate· Reported by Workspace-Bench· via BenchBoard docs47.1%
20Kimi K2.5Codex · Workspace-Bench-Lite · 100 tasks · rubric pass rate· Reported by Workspace-Bench· via BenchBoard docs46.9%
21MiniMax M2.7DeepAgent · Workspace-Bench-Lite · 100 tasks · rubric pass rate· Reported by Workspace-Bench· via BenchBoard docs45%
22Kimi K2.5OpenClaw · Workspace-Bench-Lite · 100 tasks · rubric pass rate· Reported by Workspace-Bench· via BenchBoard docs44.5%
23GPT 5.4Hermes · Workspace-Bench-Lite · 100 tasks · rubric pass rate· Reported by Workspace-Bench· via BenchBoard docs44.3%
24MiniMax M2.7OpenClaw · Workspace-Bench-Lite · 100 tasks · rubric pass rate· Reported by Workspace-Bench· via BenchBoard docs44.1%
25MiniMax M2.7Codex · Workspace-Bench-Lite · 100 tasks · rubric pass rate· Reported by Workspace-Bench· via BenchBoard docs42.7%
26Kimi K2.5DeepAgent · Workspace-Bench-Lite · 100 tasks · rubric pass rate· Reported by Workspace-Bench· via BenchBoard docs41.6%
27Qwen3.6 PlusDeepAgent · Workspace-Bench-Lite · 100 tasks · rubric pass rate· Reported by Workspace-Bench· via BenchBoard docs39.4%
28Gemini 3.1 Pro PreviewClaudeCode · Workspace-Bench-Lite · 100 tasks · rubric pass rate· Reported by Workspace-Bench· via BenchBoard docs37.5%
29Gemini 3.1 Pro PreviewDeepAgent · Workspace-Bench-Lite · 100 tasks · rubric pass rate· Reported by Workspace-Bench· via BenchBoard docs37.2%
30GPT 5.4DeepAgent · Workspace-Bench-Lite · 100 tasks · rubric pass rate· Reported by Workspace-Bench· via BenchBoard docs36.2%
31Gemini 3.1 Pro PreviewCodex · Workspace-Bench-Lite · 100 tasks · rubric pass rate· Reported by Workspace-Bench· via BenchBoard docs31.9%
32Gemini 3.1 Pro PreviewOpenClaw · Workspace-Bench-Lite · 100 tasks · rubric pass rate· Reported by Workspace-Bench· via BenchBoard docs31.6%
33Gemini 3.1 Pro PreviewHermes · Workspace-Bench-Lite · 100 tasks · rubric pass rate· Reported by Workspace-Bench· via BenchBoard docs27.1%

Not ranked

The source marks these results as descriptive, not capability scores, so we list them by name instead of ranking them.

Model or systemScore
Grok 4.3Codex · Workspace-Bench-Lite · 100 tasks · rubric pass rate· Reported by Workspace-Bench· via BenchBoard docs36.9%
Grok 4.3DeepAgent · Workspace-Bench-Lite · 100 tasks · rubric pass rate· Reported by Workspace-Bench· via BenchBoard docs13.8%
Grok 4.3Hermes · Workspace-Bench-Lite · 100 tasks · rubric pass rate· Reported by Workspace-Bench· via BenchBoard docs36.2%
Grok 4.3OpenClaw · Workspace-Bench-Lite · 100 tasks · rubric pass rate· Reported by Workspace-Bench· via BenchBoard docs34.4%
Seed-2.0-CodeClaudeCode · Workspace-Bench-Lite · 100 tasks · rubric pass rate· Reported by Workspace-Bench· via BenchBoard docs42.3%
Seed-2.0-CodeDeepAgent · Workspace-Bench-Lite · 100 tasks · rubric pass rate· Reported by Workspace-Bench· via BenchBoard docs34.6%
Seed-2.0-CodeHermes · Workspace-Bench-Lite · 100 tasks · rubric pass rate· Reported by Workspace-Bench· via BenchBoard docs38.6%
Seed-2.0-CodeOpenClaw · Workspace-Bench-Lite · 100 tasks · rubric pass rate· Reported by Workspace-Bench· via BenchBoard docs40.1%
Seed-2.0-LiteClaudeCode · Workspace-Bench-Lite · 100 tasks · rubric pass rate· Reported by Workspace-Bench· via BenchBoard docs53%
Seed-2.0-LiteDeepAgent · Workspace-Bench-Lite · 100 tasks · rubric pass rate· Reported by Workspace-Bench· via BenchBoard docs36%
Seed-2.0-LiteHermes · Workspace-Bench-Lite · 100 tasks · rubric pass rate· Reported by Workspace-Bench· via BenchBoard docs45.9%
Seed-2.0-LiteOpenClaw · Workspace-Bench-Lite · 100 tasks · rubric pass rate· Reported by Workspace-Bench· via BenchBoard docs46.6%

Data credits: BenchBoard by SYuan03 (github.com/SYuan03/benchboard). MIT covers its code and data organization; no results licence is inferred. Numeric relays retain the original stating-source links.