Skip to content
VECTOR WIREAI INTELLIGENCE
⌘K
UTC
Models741Deals30d269Policy30d65, 14 pendingSources1,014

Grok4.5

SpaceXAIReleased Jul 8, 2026Current

Score basis

Grok4.5 is capable in long context, factuality, reasoning, multimodal tasks, agentic tasks, and coding; and behind the leaders in instruction following and math. Too few results yet to rate safety or multilingual tasks.

Capability profile

Bars show the model's median result as a share of the leading model's, per capability. Select a row to see its results.

Capable6 capabilities
  1. Long Context−10.5%2 of 3
  2. Factuality−13.5%3 of 4
  3. Reasoning−13.7%6 of 6
  4. Multimodal−21.8%2 of 6
  5. Agentic−22.5%4 of 7
  6. Coding−24.2%6 of 10
Limited2 capabilities
  1. Instruction Following−36.7%1 of 3
  2. Math−41.7%5 of 5
Not rated2 capabilities

Too few results yet to rate Safety or Multilingual.

Price

$2.00input$6.00outputper million tokens

From Artificial Analysis · 3 providers tracked · All prices

Costs more than 75% of 330 priced models · 3:1 input-to-output blend, log scale

Evidence

56results on47benchmarks

  • 22 independently verified
  • 16 aggregator
  • 12 vendor-reported
  • 6 cross-referenced

From 15 sources · latest Oct 8, 2026 · How verification works

API features

Tool callingStructured outputsReasoning

As listed by OpenRouter

Research

4 papers reference Grok4.5

Grok4.5 benchmark results

56 results on 47 benchmarks, grouped by capability. Each bar is the result as a share of the capability leader's; every result links to its source.

Long Context

CapableFull long context ranking

10.5% behind the leader2 of 3 ranked benchmarks measured

Factuality

CapableFull factuality ranking

13.5% behind the leader3 of 4 ranked benchmarks measured

Reasoning

CapableFull reasoning ranking

13.7% behind the leader6 of 6 ranked benchmarks measured

Show 5 more reasoning resultsHide 5 reasoning results

Multimodal

CapableFull multimodal ranking

21.8% behind the leader2 of 6 ranked benchmarks measured

Show 1 more multimodal resultHide 1 multimodal result

Agentic

CapableFull agentic ranking

22.5% behind the leader4 of 7 ranked benchmarks measured

Show 1 more agentic resultHide 1 agentic result

Coding

CapableFull coding ranking

24.2% behind the leader6 of 10 ranked benchmarks measured

Show 2 more coding resultsHide 2 coding results

Instruction Following

LimitedFull instruction following ranking

36.7% behind the leader1 of 3 ranked benchmarks measured

Math

LimitedFull math ranking

41.7% behind the leader5 of 5 ranked benchmarks measured

More results

Benchmarks outside the capability baskets. They are not ranked against a leader.

Show 15 more resultsHide 15 results

Grok4.5: common questions

Who makes Grok4.5?

Grok4.5 is made by SpaceXAI.

When was Grok4.5 released?

Grok4.5 was released on Jul 8, 2026, according to Artificial Analysis.

What is Grok4.5 good at?

Grok4.5 is capable in long context, factuality, reasoning, multimodal tasks, agentic tasks, and coding; and behind the leaders in instruction following and math. Too few results yet to rate safety or multilingual tasks.

How much does Grok4.5 cost?

Grok4.5 costs $2.00 per million input tokens and $6.00 per million output tokens, according to Artificial Analysis. We track its price at 3 providers. At a mix of three input tokens to one output token, it costs more than 75% of the 330 priced models we track.

How many benchmarks has Grok4.5 been tested on?

We track 56 results for Grok4.5 on 47 benchmarks from 15 sources, 22 of them independently verified. The latest was recorded on Oct 8, 2026.

Which API features does Grok4.5 support?

OpenRouter lists tool calling, structured outputs, and reasoning for Grok4.5.

About this record

Where Grok4.5's numbers come from, and every name it appears under.

Tracked since
Jun 29, 2026
Newest source mention
Sep 26, 2026

Where the results come from

Verification: 56 scores · 22 independently verified · 16 aggregator-attributed · 6 vendor cross-reference · 12 vendor-reported. How these tiers are assigned

From 15 sources on 10 sites. Artificial Analysis supplies 16 of them; the 22 independently verified results come from 6 sites. Bars are coloured by trust tier.

  • artificialanalysis.ai16
  • api.llm-stats.com8
  • arcprize.org8
  • livebench.ai7
  • deepmind.google6
  • x.ai4
  • epoch.ai3
  • datasets-server.huggingface.co2
  • lmarena.ai1
  • simple-bench.com1

Also known as

How our sources name Grok4.5 at each reasoning setting.

SettingShort form
lowgrok 4.5 (low)
mediumgrok 4.5 (medium)
highgrok 4.5 (high) grok 4.5 high
Also listed asgrok 4.5grok-latestgrok-4-5