Compare AI models

Specifications
Parameters
—1.6T
Context window
1M1M
API pricing
Input price
$0.10$0.66
Output price
$0.50$1.98
Cached input price
$0.01$0.022
Cheapest input
$0.10Amazon Bedrock$0.2066Relace
Cheapest output
$0.50Amazon Bedrock$0.689Baidu
$45.00Anthropic list price$80.39Baidu · fp8
Benchmarks
BullshitBench v2
57%35%
Terminal-Bench 4.0
39.2%6.67%
Humanity's Last Exam · with tools
57.4%60%
ARC-AGI-2
61.25%61.3%
Overview
CompanyAnthropicDeepSeek
Release dateOct 7 2026Aug 13 2026
AccessClosedOpen Weight
Model detailsView modelView model

Frequently asked questions

Claude Haiku 5.5 and DeepSeek-V4-Pro-0813 are evenly matched across the 4 benchmarks they both report (BullshitBench v2, Terminal-Bench 4.0, Humanity's Last Exam, ARC-AGI-2). Claude Haiku 5.5 is cheaper on both input and output: $0.10 vs $0.66 per million input tokens, and $0.50 vs $1.98 per million output tokens. Figures are base-tier rates. DeepSeek-V4-Pro-0813 shipped 55 days before Claude Haiku 5.5, so benchmark comparisons should account for the intervening progress.

Context windows are 1M (Claude Haiku 5.5) vs 1M (DeepSeek-V4-Pro-0813). Claude Haiku 5.5 is closed, while DeepSeek-V4-Pro-0813 is open weight.

On BullshitBench v2, Claude Haiku 5.5 leads at 57% vs DeepSeek-V4-Pro-0813 at 35%. On Terminal-Bench 4.0, Claude Haiku 5.5 leads at 39.2% vs DeepSeek-V4-Pro-0813 at 6.67%. On Humanity's Last Exam · with tools, DeepSeek-V4-Pro-0813 leads at 60% vs Claude Haiku 5.5 at 57.4%. On ARC-AGI-2, DeepSeek-V4-Pro-0813 leads at 61.3% vs Claude Haiku 5.5 at 61.25%.