Announcing Artificial Analysis Intelligence Index v4.3

Announcing Artificial Analysis Intelligence Index v4.3, upgrading Terminal-Bench to v4 and adding AutomationBench-AA, an agentic workflow automation benchmark with a private test set. This is a continuation of our rollout of Intelligence Index v5

Changelog (Index v4.2 → Index v4.3):

→ Terminal-Bench: v2.1 → v4, completing our upgrade to the latest version of Terminal-Bench

→ Replacing 𝜏³-Banking with AutomationBench-AA, our implementation of Zapier’s business workflow automation benchmark

We are continuing to prioritize keeping Intelligence Index as useful as possible by bringing forward a subset of the changes we had planned for Index v5. Each change in v4.2 and v4.3 stands on its own merits and brings the index closer to real-world problem solving, adds more private test sets to prevent gaming, and reduces saturation.

Intelligence Index v4.3 raises the difficulty of agentic coding tasks and broadens the types of agentic workflows tested. Because we use a held-out test set for AutomationBench in collaboration with Zapier , the weight assigned to evaluations with private tasks or answers increases from 40% to 45%. Category weights are unchanged from v4.2: Agents 30%, Coding 20%, General 30%, Scientific Reasoning 20%.

➤ Upgraded Terminal-Bench v2.1 to v4.0: 66 multi-step tasks testing agents on tasks run in agent sandboxes driven via the terminal, including tasks involving software engineering, machine learning, science, and operations. The v4.0 update recalibrates compute and time allowances, and improves task instructions and verification. We have changed from the Terminus 2 harness to mini-SWE-agent, a minimal, model-agnostic harness. We will also be updating our Coding Agent Index, where we test model and harness pairs, to include Terminal-Bench v4.0 soon

➤ Replaced 𝜏³-Banking with AutomationBench-AA: Our implementation of Zapier’s AutomationBench tests agents on 657 business workflows across simulated applications such as Gmail, Slack, Salesforce, and Jira. Agents must complete task objectives while following business rules. AutomationBench-AA uses Zapier’s private set of 657 tasks, and is built on v1.0.6

➤ Claude Fable 5.1 and GPT-6 Astra lead the Intelligence Index: Both Claude Fable 5.1 (max with fallback) and GPT-6 Astra (max) score 53 on Intelligence Index v4.3, followed by Claude Opus 5 (max, 51), Claude Fable 5 (with fallback, 50), Muse Spark 1.3 (max, 48) and GPT-5.6 Sol (max, 47)

➤ GLM-5.3 and Kimi K3 continue to lead open weights models (both at 44): GLM-5.3 Flash (42) is the third strongest open weights model, followed by Qwen3.8 2.4T A95B (40) and DeepSeek V4 Pro 0813 (max, 36)

➤ 4 labs occupy the Intelligence vs. Cost per Task Pareto frontier: OpenAI occupies the majority of the cost-efficiency frontier, with all five reasoning efforts of the recently released GPT-6 Astra offering the lowest Cost per Task at their respective levels of intelligence. Claude Fable 5.1 (xhigh, max, 53), GLM-5.3-Flash (42) and MiMo-V2.5-Pro (26) round out the rest of the frontier

We are upgrading from Terminal-Bench v2.1 to v4.0, which features harder tasks from a terminal. Terminal-Bench v4.0 tests whether an agent can complete complex work through the terminal, across software, machine learning, science, operations, security, hardware, and media. The update recalibrates compute and time allowances and improves task instructions, environments, and verification. We run all 66 tasks three times and report average pass@1. We use mini-SWE-agent, a minimal, model-agnostic harness, for all runs.

GPT-6 Astra (max) scores 59.1%, compared with 52.0% for Claude Fable 5.1 (max with fallback) and 49.0% for Claude Opus 5 (max). Astra is 19 percentage points ahead of GPT-5.6 Sol (max), which scores 39.9%.

We are replacing 𝜏³-Banking with AutomationBench-AA, featuring broader business workflows across applications.

In collaboration with Zapier, we run the held-out test set of 657 tasks, using the v1.0.6 version of the benchmark. We call our implementation AutomationBench-AA because we award partial credit for completed objectives, with any guardrail violation reducing the task’s score to zero.

Its 657 tasks span Finance, HR, Marketing, Operations, Sales, and Support. Agents work across simulated business applications and discover the relevant APIs to complete each task. For each task, we measure the share of objectives completed. Any guardrail violation gives that task a score of zero. ‘Score’ averages these task scores across all 657 workflows and is the metric used in the Intelligence Index. ‘Tasks Completed’ separately reports the share of workflows where every objective is completed without a guardrail violation.

GPT-6 Astra (max) scores 68.5%, compared with 66.7% for Grok 4.6 (high) and 62.2% for GLM-5.3 (max). Astra (max) completes every objective without a guardrail violation on 41.6% of workflows, compared with 32.1% for Claude Fable 5.1 (max with fallback) and 28.3% for Claude Opus 5 (max). Completing every objective while respecting all guardrails remains harder than completing part of a workflow.

GLM-5.3 and Kimi K3 continue to lead open weights models at 44, followed by GLM-5.3 Flash (42), Qwen3.8 2.4T A95B (40) and DeepSeek V4 Pro 0813 (max, 36). That puts both leading open weights models 9 points behind Claude Fable 5.1 and GPT-6 Astra, which both score 53

Similar Intelligence Index scores can come at very different costs. GPT-6 Astra (max) and Claude Fable 5.1 (max with fallback) both score 53, but their average cost per Intelligence Index task is $3.26 and $7.63 respectively - 57% lower for Astra.

GLM-5.3 Flash and GPT-5.6 Terra (max) both score 42 on the Intelligence Index, but GLM-5.3 Flash costs just 18% as much per task ($0.25 versus $1.40). At a lower price point, GPT-5.6 Luna (max) scores 38 at $0.18 per task.

As always, our full methodology is available: https://artificialanalysis.ai/methodology

Category contributions to the Intelligence Index remain at Agents: 30%, Coding: 20%, General: 30%, and Scientific Reasoning: 20%. Terminal-Bench 4.0 keeps the same weighting as Terminal-Bench 2.1, and AutomationBench-AA replaces 𝜏³-Banking at its 5% weighting.

Evaluations with private questions or answers account for 45% of the Intelligence Index v4.3 weighting, up from 40% in v4.2.

Explore the full results on Artificial Analysis: https://artificialanalysis.ai/evaluations/artificial-analysis-intelligence-index