Compare

DeepSeek R2 vs Claude Haiku 4.5

DeepSeek vs Anthropic. Specs, benchmarks, and real per-task cost — all in one page.

Verdict

DeepSeek R2 is ~1.7× cheaper on a 3:1 input:output blend. DeepSeek R2 is open-weights, the other is proprietary.

DeepSeek

DeepSeek’s reasoning variant. Competes with o4 on math at a fraction of the cost.

Frontier math at open-weights pricingLong CoT traces

Anthropic

Fast, cheap, and smart enough for most routing and extraction tasks. Great base model for subagents.

Sub-second latency10x cheaper than OpusStrong structured output

Pricing

Input / 1M$0.55$0.80

Output / 1M$2.19$4.00

Context128K200K

Max output64K16K

LicenseOpen weightsProprietary

Released2026-01-152025-10-01

Benchmarks

LMSYS ELO—1305.0

MMLU Pro87.881.4

HumanEval90.187.0

SWE-Bench—49.2

MATH94.176.8

GPQA71.2—

IFEval—87.3

Per-task cost

Summarize a 1-hour meeting transcript$0.0093$0.014

Review a 500-line pull request$0.0088$0.014

Answer a customer support ticket$0.0034$0.0056

Extract structured data from a resume$0.0049$0.0080

Debug a stack trace with context$0.0077$0.013

Per-call cost using published token counts for each task. Real-world prompts vary.

Add up to four models, tweak tasks live.