Claude Haiku 5.5 vs DeepSeek V4.1 Flash 1017

Compare up to 4 models side by side.

Capability
160↑159↓ 1%
1401—
Priceper 1M tokens
Input$0.11↑ ×7.3$0.015–$0.292list $0.30-95%↓
Output$0.55↑ ×9.2$0.06–$1.168list $1.20-95%↓
Cache read$0.011↑ ×36.7$0.0003–$0.00584list $0.006-95%↓
Web searchper request$0.011↑ ×15.0$0.000732↓
Limits and modalities
Context1M1M
Max output128K↓ 67%384K↑
ModalitiesInputTextImagePDFOutputTextInputTextImageOutputText
Availability
ProvidersAnthropic, Anthropic (AWS)ByteDance (Volcano Engine), Tencent Cloud TokenHub, DeepSeek, Cloudflare Workers AI
30-day uptime——
Measured, last 30 days
Avg latency——
Effective price——
Cache hit rate——
About
Released2026-10-072026-10-17
Knowledge cutoff2026-06—

Capability scores are from Epoch AI’s Epoch Capabilities Index (CC BY 4.0, as of 2026-11-02), which combines dozens of benchmarks with GPT-5 set at 150. epoch.ai/eci

Human preference ratings are from the LMArena Text Arena (CC BY 4.0, as of 2026-10-28): Elo ratings from blind votes, on a different scale from capability scores. lmarena.ai/leaderboard/text