Skip to content

Topic

AI agent benchmarks

Evaluations that score AI systems on completing multi-step tasks end to end in a real environment, rather than on single-turn answers.

Current stories

buildOne report1 publisher

Pine AI's agent computer finishes fewer benchmark tasks than two rivals, by Pine's own count

Pine AI's cloud computer for agents resolved 27.4% of SaaS-Bench tasks in Pine's own test, below two rivals at 31.1% and 29.2%. Its higher step score and much lower token bill come from whole systems running different models, so the computer's own contribution has not been measured.

Publishers:runtimewire.com

Reality

Evidence35
Adoption10
Hype gap+40
Incentives80
Confidence40
buildOne report1 publisher

Every open-weight model KaliBench tested fell below 42% on exact-command accuracy

KaliBench, 8,504 query-command pairs across 1,642 Kali tools, found no open-weight model among 24 configurations topped 42% exact-command accuracy. Scores rise when the model is handed the tool name, but production analysts describe only intent.

Publishers:dev.to

Reality

Evidence40
Adoption
Not scored
Hype gap+10
Incentives
Not scored
Confidence40
buildOne report1 publisher

Frontier data agents average 59.5 on Argo-Bench's 235-table warehouse tasks

Frontier models average 59.5 on Argo-Bench and clear 95 on only 34.8% of its 210 enterprise data tasks. The benchmark grades the bans and refunds an agent files against a hidden simulator, a step outside what text-to-SQL scores measure.

Publishers:dev.to

Reality

Evidence35
Adoption
Not scored
Hype gap+20
Incentives
Not scored
Confidence30
investOne report1 publisher

Claude Opus 5.5 charges 5.3 times GPT-6 Sol's price per task for a nine-point benchmark lead

US companies are buying the cheapest AI model that can finish a job, the FT reports, with Ramp data pointing to a 41% fall in effective token prices. Investors now have to value AI vendors on cost per completed task, a figure that shifts with the workload.

Reality

Evidence42
Adoption50
Hype gap+20
Incentives
Not scored
Confidence40
scienceOne report1 publisher

A Berkeley scanning agent scores 100% on five AI agent benchmarks without solving a task

Berkeley's RDI center attacked the step where each benchmark computes its score, and without solving a task its own scorecard reports 100% on five of the eight, about 98% on GAIA and 73% on OSWorld.

Publishers:rdi.berkeley.edu

Reality

Evidence45
Adoption35
Hype gap+20
Incentives55
Confidence50
buildOne report1 publisher

Seven finance experts wrote the 537 SEC-filing tasks that hold o3 to 46.8 percent

The Finance Agent Benchmark scores agents on recent SEC filings using Google Search and EDGAR access, and its best reported result is OpenAI's o3 at 46.8 percent accuracy and $3.79 a query. A full pass runs about $2,035.

Publishers:arxiv.org

Reality

Evidence58
Adoption
Not scored
Hype gap+10
Incentives57
Confidence55
investOne report1 publisher

A $14.34 router matched Opus-5's score on LiteLLM's 21-task benchmark

LiteLLM's own Terminal-Bench run puts a gpt-5.4-mini classifier routing across Haiku, Sonnet and Opus at $14.34 against Opus-5's $19.74 for the same 16 solved tasks. The saving works out at 26 cents a task.

Publishers:docs.litellm.ai

Reality

Evidence45
Adoption15
Hype gap+35
Incentives80
Confidence48