Glimmer Index
Plus $4/mo

Grok 4.5

xAI · Released Jul 8, 2026 · Closed weights

xAI's Grok model for chat, coding, agentic tools, and lower hallucination risk

Strongest at
Reasoning
Top 40% of the past year's models · 4 of 5 benchmarks
Price per 1M tokens
$2 / $6
input / output · xAI
Context window
500K
tokens in · 500K out
Since the last one
82 days
after Grok 4.3
Lab announcement ↗Open in the timeline

Capability profile

Scored against 91 models released in the past year. 50 is the average model; further out is better.

This modelGrok 4.3Median of the past year
5090Median of the past yearCoding: 50 · Better than 32%Writing: 44 · Better than 19%Reading & documents: 49 · Better than 38%Reasoning: 63 · Top 40%Math: 55 · Better than 48%Science: 59 · Better than 24%Visual reasoning: 45 · Better than 29%Computer use: 60 · Better than 31%Grok 4.3: 1Grok 4.3: 29Grok 4.3: 25Grok 4.3: 22Grok 4.3: 39Grok 4.3: 52Grok 4.3: 29CodingWritingReadingReasoningMathScienceVisionComputer use
Coding50+48.9Better than 32%
Writing44+14.7Better than 19%
Reading & documents49+23.8Better than 38%
Reasoning63+41.3Top 40%
Math55+15.5Better than 48%
Science59+6.9Better than 24%
Visual reasoning45+16.0Better than 29%
Computer use60Better than 31%
Where these scores come from

Each capability combines the published benchmarks listed here, rescaled so 50 is the reference average and 15 points is one standard deviation. A hollow dot rests on a single benchmark.

Price against capability

The shaded corner holds anything better and cheaper. Prices blend input and output 3 to 1.

8 models from the past year score higher on Reasoning for less: GPT-6.1 Terra, Gemini 3.8 Flash, DeepSeek V4.1 Flash 1017 and 5 more.

$0.1$0.3$1$3$10$3020406080100GPT-6.1 Terra: 99, $2 per 1MDeepSeek V4.1 Flash 1017: 74, $0.26 per 1MClaude Haiku 5.5: 64, $0.2 per 1MMistral Large 4: 53, $2.06 per 1MGemini 4 Argon: 85, $4 per 1MGPT-6.1 Sol: 87, $4 per 1MClaude Sonnet 5.5: 81, $4 per 1MGPT-6 Sol: 76, $4 per 1MGPT-6 Luna: 53, $0.2 per 1MClaude Opus 5.5: 84, $8 per 1MGrok 4.7: 60, $3 per 1MDeepSeek V4.1 Flash: 68, $0.21 per 1MGPT-6 Astra: 90, $20 per 1MMuse Spark 1.3: 60, $2 per 1MGemini 3.8 Flash: 75, $1.5 per 1MClaude Fable 5.1: 77, $20 per 1MGLM-5.3-Flash: 48, $0.24 per 1MDeepSeek V4 Flash Vision Exp: 54, $0.32 per 1MQwen3.8 27B: 49, $1.12 per 1MGLM-5.3: 60, $2.15 per 1MGemini 3.7 Flash: 73, $1.5 per 1MDeepSeek V4 Pro 0813: 66, $0.99 per 1MGrok 4.6: 66, $3 per 1MMuse Spark 1.2: 64, $2 per 1MQwen3.8 Max: 66, $3 per 1MDeepSeek V4 Flash 0731: 61, $0.15 per 1MClaude Opus 5: 76, $10 per 1MGemini 3.6 Flash: 61, $1.5 per 1MGemini 3.5 Flash Lite: 35, $0.85 per 1MKimi K3: 62, $6 per 1MQwen3.7 Flash: 42, $0.055 per 1MGPT-5.6 Terra: 66, $4.5 per 1MGPT-5.6 Sol: 76, $8 per 1MGPT-5.6 Luna: 57, $0.45 per 1MClaude Sonnet 5: 64, $4 per 1MGLM-5.2: 49, $2.15 per 1MKimi K2.7 Code: 51, $1.71 per 1MClaude Fable 5: 74, $20 per 1MQwen3.7 Plus: 44, $0.7 per 1MClaude Opus 4.8: 68, $10 per 1MQwen3.7 Max: 58, $3.75 per 1MGemini 3.5 Flash: 64, $3.38 per 1MDeepSeek V4 Pro: 48, $1.19 per 1MDeepSeek V4 Flash: 21, $0.11 per 1MGPT-5.5 Pro: 79, $67.5 per 1MGPT-5.5: 75, $11.2 per 1MQwen3.6 27B: 27, $1.35 per 1MKimi K2.6: 46, $1.71 per 1MGrok Build 0.1: 34, $1.25 per 1MClaude Opus 4.7: 63, $10 per 1MQwen3.6 35B-A3B: 49, $0.84 per 1MMuse Spark 1.1: 61, $2 per 1MGLM-5.1: 54, $2.15 per 1MQwen3.6 Plus: 33, $1.12 per 1MGPT-5.4 nano: 48, $0.46 per 1MGPT-5.4 mini: 40, $1.69 per 1MGPT-5.4 Pro: 77, $67.5 per 1MGPT-5.4: 68, $5.62 per 1MQwen3.5 Flash: 47, $0.18 per 1MQwen3.5 122B-A10B: 44, $1.1 per 1MGemini 3.1 Pro Preview: 68, $4.5 per 1MClaude Sonnet 4.6: 54, $6 per 1MQwen3.5 Plus: 43, $0.9 per 1MQwen3.5 397B-A17B: 45, $1.35 per 1MGLM-5: 47, $1.55 per 1MClaude Opus 4.6: 55, $10 per 1MKimi K2.5: 46, $1 per 1MGLM-4.7: 51, $1 per 1MGemini 3 Flash Preview: 58, $1.12 per 1MGPT-5.2 Pro: 69, $57.8 per 1MGPT-5.2 Codex: 37, $4.81 per 1MGPT-5.2: 59, $4.81 per 1MClaude Opus 4.5: 61, $10 per 1MGPT-5.1: 54, $3.44 per 1MGrok 4.3: 22, $1.56 per 1MGrok 4.5: 63, $3 per 1MGrok 4.5Price per 1M tokens →↑ Reasoning$0.1$1$1020406080100GPT-6.1 Terra: 99, $2 per 1MDeepSeek V4.1 Flash 1017: 74, $0.26 per 1MClaude Haiku 5.5: 64, $0.2 per 1MMistral Large 4: 53, $2.06 per 1MGemini 4 Argon: 85, $4 per 1MGPT-6.1 Sol: 87, $4 per 1MClaude Sonnet 5.5: 81, $4 per 1MGPT-6 Sol: 76, $4 per 1MGPT-6 Luna: 53, $0.2 per 1MClaude Opus 5.5: 84, $8 per 1MGrok 4.7: 60, $3 per 1MDeepSeek V4.1 Flash: 68, $0.21 per 1MGPT-6 Astra: 90, $20 per 1MMuse Spark 1.3: 60, $2 per 1MGemini 3.8 Flash: 75, $1.5 per 1MClaude Fable 5.1: 77, $20 per 1MGLM-5.3-Flash: 48, $0.24 per 1MDeepSeek V4 Flash Vision Exp: 54, $0.32 per 1MQwen3.8 27B: 49, $1.12 per 1MGLM-5.3: 60, $2.15 per 1MGemini 3.7 Flash: 73, $1.5 per 1MDeepSeek V4 Pro 0813: 66, $0.99 per 1MGrok 4.6: 66, $3 per 1MMuse Spark 1.2: 64, $2 per 1MQwen3.8 Max: 66, $3 per 1MDeepSeek V4 Flash 0731: 61, $0.15 per 1MClaude Opus 5: 76, $10 per 1MGemini 3.6 Flash: 61, $1.5 per 1MGemini 3.5 Flash Lite: 35, $0.85 per 1MKimi K3: 62, $6 per 1MQwen3.7 Flash: 42, $0.055 per 1MGPT-5.6 Terra: 66, $4.5 per 1MGPT-5.6 Sol: 76, $8 per 1MGPT-5.6 Luna: 57, $0.45 per 1MClaude Sonnet 5: 64, $4 per 1MGLM-5.2: 49, $2.15 per 1MKimi K2.7 Code: 51, $1.71 per 1MClaude Fable 5: 74, $20 per 1MQwen3.7 Plus: 44, $0.7 per 1MClaude Opus 4.8: 68, $10 per 1MQwen3.7 Max: 58, $3.75 per 1MGemini 3.5 Flash: 64, $3.38 per 1MDeepSeek V4 Pro: 48, $1.19 per 1MDeepSeek V4 Flash: 21, $0.11 per 1MGPT-5.5 Pro: 79, $67.5 per 1MGPT-5.5: 75, $11.2 per 1MQwen3.6 27B: 27, $1.35 per 1MKimi K2.6: 46, $1.71 per 1MGrok Build 0.1: 34, $1.25 per 1MClaude Opus 4.7: 63, $10 per 1MQwen3.6 35B-A3B: 49, $0.84 per 1MMuse Spark 1.1: 61, $2 per 1MGLM-5.1: 54, $2.15 per 1MQwen3.6 Plus: 33, $1.12 per 1MGPT-5.4 nano: 48, $0.46 per 1MGPT-5.4 mini: 40, $1.69 per 1MGPT-5.4 Pro: 77, $67.5 per 1MGPT-5.4: 68, $5.62 per 1MQwen3.5 Flash: 47, $0.18 per 1MQwen3.5 122B-A10B: 44, $1.1 per 1MGemini 3.1 Pro Preview: 68, $4.5 per 1MClaude Sonnet 4.6: 54, $6 per 1MQwen3.5 Plus: 43, $0.9 per 1MQwen3.5 397B-A17B: 45, $1.35 per 1MGLM-5: 47, $1.55 per 1MClaude Opus 4.6: 55, $10 per 1MKimi K2.5: 46, $1 per 1MGLM-4.7: 51, $1 per 1MGemini 3 Flash Preview: 58, $1.12 per 1MGPT-5.2 Pro: 69, $57.8 per 1MGPT-5.2 Codex: 37, $4.81 per 1MGPT-5.2: 59, $4.81 per 1MClaude Opus 4.5: 61, $10 per 1MGPT-5.1: 54, $3.44 per 1MGrok 4.3: 22, $1.56 per 1MGrok 4.5: 63, $3 per 1MGrok 4.5Price per 1M tokens →↑ Reasoning

OpenRouter $2 / $6 · xAI $2 / $6 · Cached input $0.3

Since Grok 4.3

11 releases in the Grok 4.7 line.

Grok 4Jul 9, 2025131dGrok 4.1Nov 17, 2025112dGrok 4.20 (Non-Reasoning)Mar 9, 20260dGrok 4.20 (Reasoning)Mar 9, 202639dGrok 4.3Apr 17, 202682dGrok 4.5Jul 8, 202635dGrok 4.6Aug 12, 2026
Released82 days later
Price in / out$2 / $6+92%
Context500Kwas 1M
Coding50+48.9
Writing44+14.7
Reading & documents49+23.8
Reasoning63+41.3
Math55+15.5
Science59+6.9
Visual reasoning45+16.0

A newer release is out: Grok 4.7.

ECI STANDING#43 of 270
SCORE154.07152.36–156.18 CI
FRONTIER LAGLaunched 6.9 months behindScores about what GPT-5.2 Pro did in December 2025.
SCORES LIKE THESE MODELSGrok 4.7xAI · Sep 2026154.37Muse Spark 1.1Meta AI · Jul 2026154.05Qwen3.7-MaxAlibaba · May 2026154.04GPT-5.2OpenAI · Dec 2025153.66
Full Epoch Capabilities Index ranking ↗

Epoch benchmark boards

Best result on each board it has run. The tick marks the leader.

DTBench
96%#12 of 179
GPQA diamond
93%#17 of 234
SimpleBench
70%#13 of 89
OTIS Mock AIME 2024-2025
98%#33 of 216
Surface Evolver Bench
74%#8 of 26
APEX-Agents
56%#16 of 51
ARC-AGI
87%#36 of 107
ARC-AGI-2
53%#38 of 103
9 more boards
SimpleQA Verified
48%#32 of 85
WeirdML
46%#62 of 141
FrontierCode
42%#21 of 46
Chess Puzzles
35%#46 of 93
FrontierMath-Tiers-1-3-v2-Private
57%#46 of 90
ProofBench
31%#42 of 81
FrontierMath-Tier-4-v2-Private
24%#47 of 72
PostTrainBench
23%#9 of 11
Furniture Assembly
22%#31 of 35

xAI's past year

19 release days in the past year, about one every 19 days. The next is expected around Nov 10, 2026. Forecasts ↗

NovDecJanFebMarAprMayJunJulAugSepOctNov

Specifications

RELEASE DATEJul 8, 2026
WEIGHTSClosed weights
CONTEXT WINDOW500K
MAX OUTPUT500K
REASONINGNot listed
TOOL CALLINGNot listed
INPUTStext, image/jpeg, image/png
OUTPUTStext

Models.dev record ↗

Specifications and prices from Models.dev; availability and limits may differ by provider. Benchmarks from Epoch AI (CC BY 4.0), LiveBench and the other sources named above.