Skip to content
aiqbench

Anthropic · model passport

Anthropic: Claude Opus 4.6

claude-opus-4-6
  • Cloud only
  • Proprietary licence
  • 1,000,000 context
  • Released 2026-09-13(first seen on LMArena)
19/27axes covered
Overall · LMArena
1,503#5 of 396
Best percentile
99.5Instruction following
Cohort standing
78.7median 60.4

What the data says

Benchmarked on 19 of 27 tracked axes. Every figure below is the model's own published result on that board.

Strongest axes

  • 99.5 pct — Instruction following (1,512)
  • 99.5 pct — Hard prompts (1,531)
  • 99.2 pct — Coding (1,548)

Uneven: 76.6 percentile points between its best and worst axis — 99.5 pct on Instruction following versus 22.9 pct on HHEM Hallucination (Vectara).

  • Instruction following99.5 pct1,512Verify →
  • Hard prompts99.5 pct1,531Verify →
  • Coding99.2 pct1,548Verify →
  • Overall99 pct1,503Verify →
  • Vision Arena98.7 pct1,299Verify →
  • Factuality Arena97.7 pct1,493Verify →
  • Document Arena97.4 pct1,495Verify →
  • FrontierMath91.3 pct40.7%Verify →
  • Terminal-Bench90.9 pct79.8%Verify →
  • SWE-bench Verified90.6 pct78.7%Verify →
  • SimpleBench83.9 pct67.6%Verify →
  • GPQA Diamond83.3 pct90.5%Verify →
  • AIME (math)79 pct94.4%Verify →
  • Humanity's Last Exam78.3 pct34.4%Verify →
  • ARC-AGI-276.4 pct69.2%Verify →
  • WebDev Arena74.1 pct1,547Verify →
  • IFBench (Artificial Analysis)60.8 pct53.1%Verify →
  • SimpleQA Verified59.8 pct47.0%Verify →
  • HHEM Hallucination (Vectara)22.9 pct87.8%Verify →
  • Search ArenaNot benchmarked
  • Agent ArenaNot benchmarked
  • LiveBenchNot benchmarked
  • BFCL v4 (function calling)Not benchmarked
  • SWE-bench Verified (live board)Not benchmarked
  • Terminal-Bench 4.0Not benchmarked
  • MMLU-Pro (Artificial Analysis)Not benchmarked
  • LiveCodeBench (Artificial Analysis)Not benchmarked

Percentiles are the model's standing within each board's own population. Axes marked "not benchmarked" have no published result for this model — that is an absence of data, not a low score.

How to run it

Use it hosted

API model IDanthropic/claude-opus-4.6
Context length1,000,000 tokens
Max output128,000 tokens
Acceptstext, image, file
Input price$5 per 1M tokens
Cached input price$0.50 per 1M tokens
Output price$25 per 1M tokens

API parameters this endpoint supports

Include reasoningMax completion tokensMax tokensReasoningReasoning effortResponse formatStop sequencesStructured outputsTemperatureTool choiceTool callingTop-kTop-pVerbosity

Example request

curl https://openrouter.ai/api/v1/chat/completions \
  -H "Authorization: Bearer $OPENROUTER_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"anthropic/claude-opus-4.6","messages":[{"role":"user","content":"Hello"}]}'

View on OpenRouter →

Identifiers, limits and prices are synced from the upstream source shown above. Any field we haven't synced is left out entirely rather than guessed.

  • Claude Opus 4.6 (High) stands at 78.7 (raw mean 83.3) — measured on 19 of 27 axes.

    standingtemplateVerify →

  • Claude Opus 4.6 (High) stands at 78.7 versus a cohort median of 60.4 across 175 models released within 180 days — above the peer median.

    cohorttemplateVerify →

Every claim is a computed fact from the current board data, phrased by a low-temperature model call and rejected by a validation gate if any number is not in the source row. How the leaderboards work →