Skip to content
aiqbench

xAI · model passport

SpaceXAI: Grok 4.5

grok-4-5
  • Cloud only
  • Proprietary licence
  • 500,000 context
  • Released 2026-09-13(first seen on LMArena)
16/27axes covered
Overall · LMArena
1,466#46 of 396
Best percentile
93.2GPQA Diamond
Cohort standing
69.1median 60.3

What the data says

Benchmarked on 16 of 27 tracked axes. Every figure below is the model's own published result on that board.

Strongest axes

  • 93.2 pct — GPQA Diamond (93.4%)
  • 89.3 pct — Coding (1,514)
  • 89.1 pct — Instruction following (1,461)

Weakest axes

  • 13 pct — Terminal-Bench 4.0 (12.4%)

Uneven: 80.2 percentile points between its best and worst axis — 93.2 pct on GPQA Diamond versus 13 pct on Terminal-Bench 4.0.

  • GPQA Diamond93.2 pct93.4%Verify →
  • Coding89.3 pct1,514Verify →
  • Instruction following89.1 pct1,461Verify →
  • Overall88.6 pct1,466Verify →
  • Hard prompts88.6 pct1,488Verify →
  • SimpleBench86.2 pct70.0%Verify →
  • Vision Arena86 pct1,280Verify →
  • AIME (math)85.5 pct97.8%Verify →
  • Factuality Arena81.4 pct1,466Verify →
  • WebDev Arena74.8 pct1,553Verify →
  • Search Arena64.3 pct1,206Verify →
  • SimpleQA Verified63.4 pct48.3%Verify →
  • ARC-AGI-260.7 pct52.6%Verify →
  • Document Arena55.3 pct1,464Verify →
  • Agent Arena43.6 pct0.001Verify →
  • Terminal-Bench 4.013 pct12.4%Verify →
  • FrontierMathNot benchmarked
  • Humanity's Last ExamNot benchmarked
  • LiveBenchNot benchmarked
  • BFCL v4 (function calling)Not benchmarked
  • SWE-bench Verified (live board)Not benchmarked
  • MMLU-Pro (Artificial Analysis)Not benchmarked
  • LiveCodeBench (Artificial Analysis)Not benchmarked
  • IFBench (Artificial Analysis)Not benchmarked
  • HHEM Hallucination (Vectara)Not benchmarked
  • SWE-bench VerifiedNot benchmarked
  • Terminal-BenchNot 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 IDx-ai/grok-4.5
Context length500,000 tokens
Max output450,000 tokens
Acceptstext, image, file
Input price$2 per 1M tokens
Cached input price$0.30 per 1M tokens
Output price$6 per 1M tokens

API parameters this endpoint supports

Include reasoningLogprobsMax tokensReasoningReasoning effortResponse formatSeedStructured outputsTemperatureTool choiceTool callingTop logprobsTop-p

Example request

curl https://openrouter.ai/api/v1/chat/completions \
  -H "Authorization: Bearer $OPENROUTER_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"x-ai/grok-4.5","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.

  • grok-4.5 stands at 69.1 (raw mean 72.7) — measured on 16 of 27 axes.

    standingtemplateVerify →

  • grok-4.5 stands at 69.1 versus a cohort median of 60.3 across 176 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 →