Benchmark Almanac

Anthropic ECI (AECI)

Claude Opus 5.5 169.4 leads since Sep 2026 Only in system cards

Index that combines a basket of capability benchmarks into one capability score to track trends across Anthropic models.

Anthropic's adaptation of the Epoch Capabilities Index. The basket and weights are internal.

Made by Anthropic Released Apr 2026 Composite indices Tasks kept private Last in a system card Oct 2026 Last model added Oct 2026
Level
Research frontier Researchers, on open problems or tasks that take days
Scored
Checked automatically A program checks the result against an answer key, unit tests or the expected final state.
Answers
Agent task Many steps in a terminal, browser, desktop, app, game or robot.
6
Models scored in lab cards
7
System cards report it, from 1 lab

Claude Opus 5.5 leads with 169.4

Best score per model as labs report them in their own cards. Labs pick their own settings, so these compare less cleanly than independent runs.

  1. 169.4 Claude Opus 5.5
  2. 167.9 Claude Sonnet 5.5
  3. 167.1 Claude Haiku 5.5
  4. 162.1 Claude Opus 5
  5. 155.5 Claude Opus 4.8
  6. 155.1 Claude Opus 4.7
  1. 1 Claude Opus 5.5 169.4
  2. 2 Claude Sonnet 5.5 167.9
  3. 3 Claude Haiku 5.5 167.1
  4. 4 Claude Opus 5 162.1
  5. 5 Claude Opus 4.8 155.5
  6. 6 Claude Opus 4.7 155.1
#ModelScoreSettingDateSource
1 Claude Opus 5.5 169.4 Sep 2026 Model card
2 Claude Sonnet 5.5 167.9 Sep 2026 Model card
3 Claude Haiku 5.5 167.1 Oct 2026 Model card
4 Claude Opus 5 162.1 Jul 2026 Model card
5 Claude Opus 4.8 155.5 May 2026 Model card
6 Claude Opus 4.7 155.1 Apr 2026 Model card

4 models have held number one since its release in Apr 2026

Lab claims count from the day the card came out, independent runs from the model’s release.

Anthropic
Lab claims 4 leaders
Claude Opus 5.5 169.4

Scores over time

All the scores we have, one dot each, so a model tested at several settings appears more than once.

Reported by the lab in its model card
155 160 165 170 2026 155 160 165 170 2026

Labs report it only in system cards

We read the headline results of 479 launch posts from 51 labs, and none of them list this benchmark. Some system cards do, below.

7 system cards report it, from 1 lab

Besides launch posts, we read the system cards, model cards and technical reports labs publish with a release. They report far more, including safety and frontier-risk tests.

DateDocumentScoreAs the document writes itSection
7 Oct 2026 Claude Haiku 5.5 system card 167.1 Anthropic ECI (AECI), most recent fit (2.3.2) AI R&D risk
28 Sep 2026 Claude Sonnet 5.5 system card 167.9 Anthropic ECI (AECI) on the most recent fit, 167.93 (2.3.2, Figure 2.3.2.A) AI R&D risk
22 Sep 2026 Claude Opus 5.5 system card 169.4 Anthropic ECI (AECI), September 18 refit, 43 benchmarks, global 95% CI 165.23 to 177.05, local 167.99 to 171.24 (Table 2.3.5.3.A) AI R&D risk
24 Jul 2026 Claude Opus 5 system card 162.1 Anthropic ECI (AECI) point estimate, 95% CI 158.0 to 167.3, n=40 benchmarks, earlier snapshot (2.3.3) AI R&D risk
28 May 2026 Claude Opus 4.8 system card 155.5 Anthropic ECI (AECI) point estimate on a reduced set of 11 evaluations, between Opus 4.7 (154.1) and Mythos Preview (158.3) on the same set (2.3.4, Figure 2.3.4.A) AI R&D risk
16 Apr 2026 Claude Opus 4.7 system card 155.1 Anthropic ECI (AECI), Claude Opus 4.7 at 155.1 with 95% CI 152.3 to 159.0, about 1.0 above the frontier trend through Opus 4.6, printed in chart legend (2.3.7, Figure 2.3.7.A) AI R&D risk
7 Apr 2026 Claude Mythos Preview system card Anthropic ECI (AECI), no AECI value printed for Mythos Preview. Slope ratios printed in the chart legend: 1.86 (break at Opus 4.5), 2.00 (Sonnet 4.5), 4.28 (Opus 4.6) (2.3.6, Figure 2.3.6.B) AI R&D risk

More in Other

All 41 in Other
  • Personal Superintelligence Index (Meta) In use Meta's composite of seven personal and knowledge-work evals: Agents' Last Exam, AutomationBench-AA, FocusBench, JobBench, GDP.pdf, VBT and DRACO.
  • Artificial Analysis Intelligence Index v5 In use Version 5 of Artificial Analysis's index: nine evaluations in four equal categories, most with private test sets, run three times each in its own agent harness. Gemini 4 Carbon 59.0
  • Qwen in-house English automatic evaluation Only in system cards Models answer Qwen in-house English datasets on knowledge, comprehension, coding, math and related skills. Skills and weights change between reports. Qwen2-72B 72.9%
  • Recognition of model-written user turns (Anthropic) Only in system cards Models write synthetic user turns for real Claude Code transcripts, then judge which version is real. Score is accuracy above chance. Claude Mythos Preview 0.35
  • LiveBench Dropped Models answer monthly refreshed questions across math, coding, reasoning, language, instruction following, and data analysis. Scored automatically against objective ground truth. Claude Fable 5.1 85.7%
  • PhiBench Dropped Models answer original prompts that Microsoft's Phi team wrote to test a wide range of skills and reasoning. Phi-4-Reasoning-plus 74.2%