xAI: Grok 4.3
xAI: Grok 4.3 is x-ai's model. On the un-leaked meo benchmark it ranks #6 of 22 with 56.6% accuracy and an Effective Value (𝕍) of 0.0103 — earning a correct answer for about $0.0182 in 27s each.
as of 2026-06-08
Per-domain accuracy
Efficiency
- Cost per correct
- $0.0182
- Seconds per correct
- 27
- Tokens per correct
- 5,679
- Correct answers
- 111 / 251
- Total benchmark cost
- $2.02
- Avg tokens / item
- 2,511
Robustness under pressure
Susceptibility index -0.031 (Δaccuracy under threatening/emotional prompts; ≈0 = robust), sway magnitude 0.052. How we measure this →
- control
- 54%
- neutral
- 46%
- user direct
- 50%
- model direct
- 46%
- user implied
- 50%
- model implied
- 58%
- positive control
- 54%
Aggregated third-party benchmarks
Artificial Analysis (artificialanalysis.ai). Redistribution requires an AA commercial license.
Model details
- Lab
- x-ai
- Released
- 2026-04-30
- Weights
- —
- Context
- 1M
- Max output
- —
- Knowledge cutoff
- —
- Price in
- $1.25/M
- Price out
- $2.50/M
- Modality
- text+image->text
Sources: openrouter · artificial_analysis · meo
FAQ
How accurate is xAI: Grok 4.3?
xAI: Grok 4.3 scores 56.6% overall on the meo benchmark (rank #6 of 22), an un-leaked private holdout graded objective-first.
What is xAI: Grok 4.3's Effective Value?
Its Effective Value (𝕍) is 0.0103 at N=10, ω=1, δ=1.5 — a fused metric of accuracy, speed, cost, and error-cascade. 𝕍 rankings shift with task depth.
How much does xAI: Grok 4.3 cost per correct answer?
About $0.0182 per correct answer (5,679 tokens, 27 seconds each) at benchmark-time pricing via OpenRouter.
Is xAI: Grok 4.3 open weights?
Open-weights status for xAI: Grok 4.3 is not recorded.