Model
Model explorer

Grok-3 (Think)

CLOSED
SpaceXAI · Grok 3 family · released Feb 17, 2025

Reasoning "Think"/Big Brain mode of Grok 3; hit #1 on Chatbot Arena at launch (~1400 Elo, per rdworldonline.com). Deprecated May 15, 2026, redirects to Grok 4.3.

ReasoningCodingVisionFunction callingTool useAgentic
1908.8
Elo · rank #65
Parameters
Undisclosed
Active params
Undisclosed
Context
131K tokens
Architecture
Mixture-of-Experts (size undisclosed; trained on Colossus, ~10x Grok-2 compute)
License
Proprietary
Languages
—
API price (in/out)
$3 / $15
Modalities
text
Benchmark results
Bar shows position within the tracked field; marker = field best
AIMEMath93.3%#26
best: GPT-5.2 · 100.0%
Arena EloHuman preference1402#22
best: Claude Fable 5 · 1505
GPQA DiamondReasoning84.6%#57
best: GPT-5.6 · 94.6%
LiveCodeBenchCoding79.4%#49
best: DeepSeek-V4-Pro (Think Max) · 93.5%
MMLU-ProKnowledge79.9%#59
best: Claude Fable 5 · 91.5%
MMMUVision78.0%#27
best: Claude Fable 5 · 89.3%
Run it locally
Closed weights — available via API only. No local deployment.
Input / M tok
$3
Output / M tok
$15
Grok 3 family
Elo progression across releases
Grok-3 miniFeb 20251715.3Grok-3 (Standard)Feb 20251505.8Grok-3 (Think)Feb 20251908.8
Resources
DOCSdocs.x.aiAPIAvailable via SpaceXAI API
API price $3/$15 · each benchmark row carries its own source badge (see methodology)