Claude Mythos Preview vs Claude Opus 4.8: Specs & Benchmark Comparison

Claude Mythos Preview and Claude Opus 4.8 both come from Anthropic, Claude Mythos Preview was released in April 2026, and Claude Opus 4.8 followed a month later in May 2026.

The two models share 12 published benchmarks. Claude Mythos Preview leads on 11 of them, Claude Opus 4.8 on 1. The widest gaps are on SWE-Bench Multimodal, where Claude Mythos Preview scores 59.0% against 38.4%; Graphwalks BFS >128k, where Claude Mythos Preview scores 80.0% against 68.1%. Averaged across everything we track, Claude Mythos Preview sits at 85.1% and Claude Opus 4.8 at 71.0%.

Claude Opus 4.8 is the cheaper API at $5 per million input tokens and $25 per million output tokens, roughly 5 times cheaper than Claude Mythos Preview at $25 and $125. Both accept text and images as input. On tooling, only Claude Opus 4.8 supports function calling and only Claude Opus 4.8 offers structured output.

CharacteristicClaude Mythos PreviewClaude Opus 4.8
CompanyAnthropicAnthropic
Release DateApril 7, 2026May 28, 2026
Parameters
MultimodalYesYes
Context (input)1.0M
Context (output)128K
Input Price / 1M$25.00$5.00
Output Price / 1M$125.00$25.00
Average Score85.1%71.0%
Benchmarks
SWE-Bench Multimodal59.0%38.4%
Graphwalks BFS >128k80.0%68.1%
SWE-Bench Pro77.8%69.2%
Terminal-Bench 2.082.0%74.6%
Humanity's Last Exam64.7%57.9%
SWE-Bench Verified93.9%88.6%
CyberGym83.1%78.8%
OSWorld-Verified79.6%83.4%
CharXiv-R93.2%89.9%
SWE-bench Multilingual87.3%84.4%
BrowseComp86.9%84.3%
GPQA94.6%93.6%

Visual Benchmark Comparison

Claude Mythos Preview
Claude Opus 4.8
SWE-Bench Multimodal0.6 vs 0.4
0.6
0.4
Graphwalks BFS >128k0.8 vs 0.7
0.8
0.7
SWE-Bench Pro0.8 vs 0.7
0.8
0.7
Terminal-Bench 2.00.8 vs 0.7
0.8
0.7
Humanity's Last Exam0.6 vs 0.6
0.6
0.6
SWE-Bench Verified0.9 vs 0.9
0.9
0.9
CyberGym0.8 vs 0.8
0.8
0.8
OSWorld-Verified0.8 vs 0.8
0.8
0.8
CharXiv-R0.9 vs 0.9
0.9
0.9
SWE-bench Multilingual0.9 vs 0.8
0.9
0.8
BrowseComp0.9 vs 0.8
0.9
0.8
GPQA0.9 vs 0.9
0.9
0.9

Verdict

Claude Opus 4.8 leads in 2 out of 4 comparison categories.

Overall Performance

Both models show comparable average scores: Claude Mythos Preview — 0.9, Claude Opus 4.8 — 0.7.

Programming

On SWE-Bench, both models are nearly equal: Claude Mythos Preview — 0.9, Claude Opus 4.8 — 0.9.

API Cost

Claude Opus 4.8 is 5.0x cheaper: input $5.00/1M vs $25.00/1M tokens.

Recency

Claude Opus 4.8 is newer: released 5/28/2026 vs 4/7/2026.

More About These Models

Frequently Asked Questions

Which is better for coding — Claude Mythos Preview or Claude Opus 4.8?
On the SWE-Bench benchmark, Claude Mythos Preview shows a better result: 93.9% vs 88.6%.
Which model is cheaper — Claude Mythos Preview or Claude Opus 4.8?
Claude Opus 4.8 is cheaper for input: $5.00 per 1M tokens vs $25.00.
Which has a larger context window — Claude Mythos Preview or Claude Opus 4.8?
Context window data is available on the individual model pages.

The Claude Mythos Preview and Claude Opus 4.8 comparison is updated for 2026. Data includes benchmark results, API pricing, context window size and other specifications. For more detailed information, visit the Claude Mythos Preview or Claude Opus 4.8 page. See also the complete list of AI model comparisons.