Claude Opus 4.8 vs Grok 4

Benchmarks, API pricing and specs, head to head. Data updated 2026-06-10.

Claude Opus 4.8

Anthropic · May 2026

Anthropic's flagship Opus with adaptive thinking, effort controls and a 1M context — 88.6% on SWE-bench Verified at $5/$25.

Grok 4

xAI · Jul 2025

xAI's mid-2025 flagship, trained on the 200K-GPU Colossus cluster — a math and science reasoning standout of its generation.

The verdict

Claude Opus 4.8 wins 1 of the 1 benchmarks these models share, against 0 for Grok 4. Grok 4 is about 1.7x cheaper per blended million tokens (3:1 input:output mix). Claude Opus 4.8 also takes 1M of context versus 256K for Grok 4.

Benchmark head-to-head 10

GPQA Diamond
~91%87.5%
Claude Opus 4.8Grok 4

Specs & pricing

Claude Opus 4.8Grok 4
modhub Index
Input price / 1M$5$3
Output price / 1M$25$15
Context window1M256K
Max output64K
Open weightsnono
Reasoning modelyesyes
Multimodal inputtext, imagetext, image
Knowledge cutoffMar 2026Jul 2025
ReleasedMay 2026Jul 2025
Example monthly cost*$87.50$52.50

* 10M input + 1.5M output tokens per month at list prices, no caching. Green = better value on that row.

Frequently asked questions

Which is better, Claude Opus 4.8 or Grok 4?
Claude Opus 4.8 wins 1 of the 1 benchmarks these models share, against 0 for Grok 4. Grok 4 is about 1.7x cheaper per blended million tokens (3:1 input:output mix). Claude Opus 4.8 also takes 1M of context versus 256K for Grok 4.
Which is cheaper, Claude Opus 4.8 or Grok 4?
Claude Opus 4.8 costs $5/$25 per million input/output tokens, while Grok 4 costs $3/$15. For a typical workload of 10M input and 1.5M output tokens per month, that's $87.50 versus $52.50.
Which model is better for coding, Claude Opus 4.8 or Grok 4?
We don't yet track SWE-bench Verified results for both models; check their individual pages for coding-related scores.

More comparisons