Claude 5.5 SonnetvsClaude 5.5 Haiku

Claude 5.5 Sonnet leads the overall Podium Score by 7.3 points (94.8 vs 87.5).Category wins: Claude 5.5 Sonnet 7 — 0 Claude 5.5 Haiku.Claude 5.5 Haiku is the cheaper pick ($4/M vs $15/M per 1M output tokens). Claude 5.5 Haiku is faster (145 t/s vs 94 t/s).

Claude 5.5 SonnetDrag slider or hover to compareClaude 5.5 Haiku
Claude 5.5 Haiku
87.5
Coding: 84.0 · Speed: 145 t/s
Output Price: $4/M
Claude 5.5 Sonnet
94.8
Coding: 91.5 · Speed: 94 t/s
Output Price: $15/M
Quick VerdictAggregated 2026 Benchmark Analysis

Claude 5.5 Sonnet is rated higher overall with a Podium Score of 94.8 (vs 87.5 for Claude 5.5 Haiku). In head-to-head domain evaluations, Claude 5.5 Sonnet outperforms in 7 of 7 benchmark categories.

Coding & Engineering
Claude 5.5 Sonnet (91.5)
Inference Speed
Claude 5.5 Haiku (145 t/s)
Cost Efficiency
Claude 5.5 Haiku ($4/M)
API Price / 1M (in · out)
Claude 5.5 Sonnet $3/M · $15/M — Claude 5.5 Haiku $0.8/M · $4/M

Choose Claude 5.5 Sonnet for long documents and large codebases, and Claude 5.5 Haiku for cost-sensitive workloads.

MetricClaude 5.5 SonnetClaude 5.5 Haiku
Podium Score94.8 ▲87.5
Arena Elo1520 ▲1410
Intelligence Index95.0 ▲88.0
Coding91.5 ▲84.0
Math94.2 ▲86.5
Reasoning95.0 ▲88.0
Agentic89.4 ▲78.5
Knowledge88.0 ▲81.0
Multimodal96.0 ▲88.0
Long-context98.0 ▲91.0
Output speed94 t/s145 t/s ▲
Output price$15/M$4/M ▲
Context window1000K ▲500K
0 / 4