Grok 4.5 Tops Agent Test, Backing Musk’s Opus-Class Claim

Elon Musk referred to as Grok 4.5 an Opus-class mannequin that runs sooner and prices much less. An impartial, agentic benchmark now supplies actual backing for that declare.

On Artificial Analysis’s AutomationBench-AA, Grok 4.5 ranked first with a 51.4% rating whereas costing $0.34 per process. It beat each Claude Fable 5 (48.6%) and Claude Opus 4.8 (48.5%).

An Independent Check on Elon Musk’s Claim

SpaceXAI took Grok 4.5 public this week, constructed on its 1.5 trillion-parameter V9 basis. Musk’s pitch rested on early inner evaluations.

AutomationBench-AA modifications that. Artificial Analysis runs the benchmark independently and retains its process set personal to forestall contamination.

The check spans 657 duties throughout 40 simulated apps, together with Gmail, Slack, Salesforce, and HubSpot. It scores the share of targets an agent completes with out breaking guardrails.

Follow us on X to get the newest information because it occurs

Grok 4.5: Cheaper, Faster, and Mostly Compliant

Grok 4.5’s $0.34 per process sat far under Fable 5’s $1.35 and Opus 4.8’s $1.46. Gemini 3.5 Flash got here closest at $0.49.

The mannequin used about 8,000 output tokens per process, roughly 1 / 4 of Opus 4.8’s whole. That effectivity drives a lot of the price hole, matching the framing Musk used at launch.

“Its whole token utilization of 0.44M per process is among the many lowest on the leaderboard. Low price is pushed by this effectivity in addition to low token pricing,” Artificial Analysis said.

In addition, Grok 4.5 completed 79.9% of task targets and totally handed 21.9% of duties. In Finance, the toughest area, it led with 71%, forward of Fable 5’s 64% and Opus 4.8’s 62%.

However, the mannequin broke extra guidelines than its closest rivals. It logged 0.63 guardrail violations per process, above Opus 4.8’s 0.55 and Gemini 3.5 Flash’s 0.46.

That hole issues for corporations that deploy brokers to reside monetary techniques, the place a single violation can incur actual prices.

Subscribe to our YouTube channel to look at leaders and journalists present knowledgeable insights

The publish Grok 4.5 Tops Agent Test, Backing Musk’s Opus-Class Claim appeared first on BeInCrypto.