SI Index Data as of Oct 8, 2026

Grok 4.20 (Reasoning): Price, Context, Benchmarks, and Release Details

xAI provisional listing · first seen Oct 8, 2026 · score computed Oct 8, 2026, 22:13 UTC
57.7
SI Score (method si-v2-absolute-shrinkage-1)
#65 of 115 ranked
80% confidence 80 percent, High confidence — 3 of 7 expected sources in
Coverage 64% of expected source weight · 100% confidence at 80% coverage

Pillar breakdown

Coding (weight 40 percent) —
Math (weight 15 percent) 43.2
Preference (weight 15 percent) 77.8
Reasoning (weight 30 percent) 65.2

Pillar weights: reasoning 30% · math 15% · coding 40% · preference 15%. Benchmark results use fixed 0–100 scales before averaging; incomplete evidence is shrunk toward 50.

Facts

Input price / 1M
not yet reported
Output price / 1M
not yet reported
Context window
1Mmodels.devPublished source fact Retrieved Oct 8, 2026 · MIT
Open source ↗
Max output
30Kmodels.devPublished source fact Retrieved Oct 8, 2026 · MIT
Open source ↗
Released
Mar 9, 2026models.devPublished source fact Retrieved Oct 8, 2026 · MIT
Open source ↗
Open weights
Nomodels.devPublished source fact Retrieved Oct 8, 2026 · MIT
Open source ↗
License
not yet reported
Input modalities
text, image, pdfmodels.devPublished source fact Retrieved Oct 8, 2026 · MIT
Open source ↗

Benchmark results

10 results
Benchmark Raw result Normalized (0–100) Pillar
arc agi v1 public eval 95.5%ARC PrizeARC Prize steward-published result; exact edition/subset; publication time is HTTP Last-Modified of the aggregate export, not the evaluation date. [variant] grok-4.20-beta-0309b-reasoningPublished Oct 6, 2026 Retrieved Oct 8, 2026 · factual citation
Open source ↗ 		95.5 reasoning
arc agi v1 semi private 89.5%ARC PrizeARC Prize steward-published result; exact edition/subset; publication time is HTTP Last-Modified of the aggregate export, not the evaluation date. [variant] grok-4.20-beta-0309b-reasoningPublished Oct 6, 2026 Retrieved Oct 8, 2026 · factual citation
Open source ↗ 		89.5 reasoning
arc agi v2 public eval 63.6%ARC PrizeARC Prize steward-published result; exact edition/subset; publication time is HTTP Last-Modified of the aggregate export, not the evaluation date. [variant] grok-4.20-beta-0309b-reasoningPublished Oct 6, 2026 Retrieved Oct 8, 2026 · factual citation
Open source ↗ 		63.6 reasoning
arc agi v2 semi private 65.1%ARC PrizeARC Prize steward-published result; exact edition/subset; publication time is HTTP Last-Modified of the aggregate export, not the evaluation date. [variant] grok-4.20-beta-0309b-reasoningPublished Oct 6, 2026 Retrieved Oct 8, 2026 · factual citation
Open source ↗ 		65.1 reasoning
arc agi v3 semi private 0.1%ARC PrizeARC Prize steward-published result; exact edition/subset; publication time is HTTP Last-Modified of the aggregate export, not the evaluation date. [variant] xai-grok-4-20-beta-0309-reasoningPublished Oct 6, 2026 Retrieved Oct 8, 2026 · factual citation
Open source ↗ 		0.1 reasoning
frontiermath tier 4 v2 17.1%Epoch AI BenchmarkingEpoch-owned evaluation mean score; scores only, no benchmark questions [variant] Published Jul 13, 2026 Retrieved Oct 8, 2026 · CC-BY
Open source ↗ 		17.1 math
frontiermath tiers 1 3 v2 44.9%Epoch AI BenchmarkingEpoch-owned evaluation mean score; scores only, no benchmark questions [variant] Published Jul 13, 2026 Retrieved Oct 8, 2026 · CC-BY
Open source ↗ 		44.9 math
gpqa diamond 89.3%Epoch AI BenchmarkingEpoch-owned evaluation mean score; scores only, no benchmark questions [variant] Published Jul 13, 2026 Retrieved Oct 8, 2026 · CC-BY
Open source ↗ 		89.3 reasoning
lmarena text 1450.8 eloLMArena / ArenaPublished source fact [variant] text / overallPublished Oct 2, 2026 Retrieved Oct 8, 2026 · CC-BY-4.0
Open source ↗ 		77.8 preference
otis mock aime 2024 2025 92.2%Epoch AI BenchmarkingEpoch-owned evaluation mean score; scores only, no benchmark questions [variant] Published Jul 13, 2026 Retrieved Oct 8, 2026 · CC-BY
Open source ↗ 		92.2 math

Normalization uses fixed absolute 0–100 scales for each unit, independently of other models that have a result on each benchmark. Coverage and evidence breadth still affect the composite; compare the evaluation conditions before reading a small score gap as decisive. “lab-reported” marks the provider's own published figure.

Sources: in and pending

in Reported (3)

  • ARC Prize · arrived Oct 8, 2026
  • Epoch AI Benchmarking · arrived Oct 8, 2026
  • LMArena / Arena · arrived Oct 8, 2026

pending Awaiting (4)

  • Humanity’s Last Exam · carries 13% of expected weight
  • Official model cards via models.dev · carries 4% of expected weight
  • LiveBench · carries 13% of expected weight
  • Terminal-Bench · carries 6% of expected weight

The confidence % rises as pending sources publish. Some sources never cover some models — that is why 100% confidence arrives at 80% of expected weight, not at full coverage.