MMODELYST
Models/Grok 4.5
MDL

Grok 4.5

SpaceXAI

Verified against Artificial Analysis · Jul 13, 2026
Capability
96
percentile index
Price /M
$2 / $6
in / out
Context
500K
tokens
Avg score
63.8
4 benchmarks
AA Intelligence
53.8
index
Speed
110 tok/s
TTFT 13s
Released
Jul 8, 2026
About

Grok 4.5 is a proprietary model in the Grok family from SpaceXAI. Benchmarked on 4 evals, averaging 63.8.

Benchmark scores · 4
avg 63.8 — higher is better
GPQA Diamond
93.1
AA Long-Context Reasoning
67.7
SciCode
54.1
Humanity's Last Exam
40.3
Details
PublisherSpaceXAI
FamilyGrok
Parameters
License
WeightsClosed
ReleasedJul 8, 2026
Performance & price
Output speed110 tok/s
Latency (TTFT)13s
Price in / out$2 / $6 per 1M
Blended price$3 / 1M
Coding index72.4

Median across API providers, via Artificial Analysis. Blended = 3:1 input:output. For reasoning models, latency includes thinking time. Methodology →

AA Long-Context Reasoningfirst seen at 67.7
4d ago
SciCodefirst seen at 54.1
4d ago
Humanity's Last Examfirst seen at 40.3
4d ago
GPQA Diamondfirst seen at 93.1
4d ago

Append-only ledger — every observed change to this model's numbers.

API providers
Artificial Analysis$6/M