Benchmarks for what frontier AI hasn't solved
Digital Work Index
An economically weighted measure of model performance across eight challenging benchmarks covering software engineering, logical reasoning and writing, and professional environments.
28.31
Gemini 4 Argon
26.4
Muse Wonder
23.62
Fable 5.1
23.61
Opus 5.5
23.03
GPT-6 Astra
21.75
Opus 5
21.19
GPT-6.1 Sol
20.52
Grok 4.6
19.3
Grok 4.7
16.92
Gemini Flash 3.8
16.76
Kimi K3
16.45
GLM 5.3
16.29
Muse Spark 1.3
11.81
Qwen 3.8 Max
11.07
DeepSeek V4 Pro
6.95
Nemotron 3 Ultra 550B
All Models
Meta
Oct 12, 2026
Muse Wonder
RANK BY BENCHMARK
Agentic Coding 2.0
43.6%
#2/16
Agents’ Last Exam v1.1
36.9%
#4/9
SnorkelFinance 2.0
15%
#9/16
SnorkelLegal
38.3%
#3/16
SnorkelManufacturing
27.8%
#1/16
Google DeepMind
Sep 30, 2026
Gemini 4 Argon
RANK BY BENCHMARK
Agentic Coding 2.0
41.8%
#4/16
Agents’ Last Exam v1.1
37.7%
#3/9
SnorkelFinance 2.0
15.1%
#8/16
SnorkelLegal
46%
#1/16
SnorkelManufacturing
19.5%
#3/16
OpenAI
Sep 29, 2026
GPT-6.1 Sol
RANK BY BENCHMARK
Agentic Coding 2.0
42.8%
#3/16
SnorkelFinance 2.0
6.6%
#14/16
SnorkelLegal
20.7%
#16/16
SnorkelManufacturing
19.6%
#2/16
SnorkelRevOps
12.1%
#9/16
Anthropic
Sep 22, 2026
Opus 5.5
RANK BY BENCHMARK
Agentic Coding 2.0
40.4%
#5/16
Agents’ Last Exam v1.1
42.6%
#1/9
SnorkelFinance 2.0
18.6%
#5/16
SnorkelLegal
31.4%
#9/16
SnorkelManufacturing
10.4%
#9/16
All benchmarks
All Benchmarks
Sort: Newest
Benchmark Type: All
Software Engineering
Open Benchmarks Grants
Senior SWE-bench
Evaluates coding agents on senior-level software engineering tasks.
Tasteful Solve Rate
Opus 5.5
39.3%
2
GPT-6 Astra
36.8%
3
Fable 5.1
34.7%
Software Engineering
Open Benchmarks Grants
LibraryDesignBench
Measures how well AI agents design libraries that other agents can use.
Pass Rate
1
Opus 5.5
48.9%
2
Fable 5.1
47.5%
3
GPT-6 Astra (mini-SWE)
45.1%
Software Engineering
Proprietary
SWE-bench CLI
Evaluates validated multi-file code changes in open-source repositories.
By pass@1
1
Opus 5.5
16.6%
2
Gemini 4 Argon
16.3%
3
Fable 5.1
14.5%
Software Engineering
Open Benchmarks Grants
SlopCode Bench
Measures coding-agent performance across evolving software requirements.
By Iso Solve
1
GPT-5.5
28.06%
2
GPT-5.3
26.02%
3
GPT-5.4
23.47%
View archived benchmarks
