Claude Mythos PreviewvsGPT-6 Luna (low)

Claude Mythos Preview leads the overall Podium Score by 22.9 points (97.4 vs 74.5).Category wins: Claude Mythos Preview 4 — 0 GPT-6 Luna (low).Claude Mythos Preview is the cheaper pick ($15/M vs $20/M per 1M output tokens). GPT-6 Luna (low) is faster (150 t/s vs 80 t/s).

Claude Mythos PreviewDrag slider or hover to compareGPT-6 Luna (low)
GPT-6 Luna (low)
74.5
Coding: 55.0 · Speed: 150 t/s
Output Price: $20/M
Claude Mythos Preview
97.4
Coding: 80.9 · Speed: 80 t/s
Output Price: $15/M
Quick VerdictAggregated 2026 Benchmark Analysis

Claude Mythos Preview is rated higher overall with a Podium Score of 97.4 (vs 74.5 for GPT-6 Luna (low)). In head-to-head domain evaluations, Claude Mythos Preview outperforms in 4 of 7 benchmark categories.

Coding & Engineering
Claude Mythos Preview (80.9)
Inference Speed
GPT-6 Luna (low) (150 t/s)
Cost Efficiency
Claude Mythos Preview ($15/M)
API Price / 1M (in · out)
Claude Mythos $3/M · $15/M — GPT-6 Luna (low) $5/M · $20/M

Choose Claude Mythos Preview for knowledge and factual QA, and GPT-6 Luna (low) for low-latency applications.

MetricClaude Mythos PreviewGPT-6 Luna (low)
Podium Score97.4 ▲74.5
Arena Elo——
Intelligence Index——
Coding80.9 ▲55.0
Math100.0—
Reasoning98.8 ▲80.0
Agentic——
Knowledge100.0 ▲45.0
Multimodal100.0 ▲85.0
Long-context——
Output speed80 t/s150 t/s ▲
Output price$15/M ▲$20/M
Context window1000K1000K
0 / 4