Compare
DeepSeek R2 vs Gemini 2.5 Ultra
DeepSeek vs Google DeepMind. Specs, benchmarks, and real per-task cost — all in one page.
Verdict
DeepSeek R2 is ~10.9× cheaper on a 3:1 input:output blend. DeepSeek R2 is open-weights, the other is proprietary.
DeepSeek
DeepSeek R2DeepSeek’s reasoning variant. Competes with o4 on math at a fraction of the cost.
Frontier math at open-weights pricingLong CoT traces
Google DeepMind
Gemini 2.5 Ultra2M-token context, native video understanding, and Google’s deepest multimodal stack. The long-context king.
2M context windowNative videoBest multimodal reasoning
Pricing
Input / 1M$0.55$7.00
Output / 1M$2.19$21.00
Context128K2.0M
Max output64K64K
LicenseOpen weightsProprietary
Released2026-01-152026-02-05
Benchmarks
LMSYS ELO—1385.0
MMLU Pro87.890.4
HumanEval90.190.2
SWE-Bench—58.3
MATH94.192.0
GPQA71.2—
MMMU—82.1
IFEval—89.7
Per-task cost
Summarize a 1-hour meeting transcript$0.0093$0.115
Review a 500-line pull request$0.0088$0.098
Answer a customer support ticket$0.0034$0.038
Extract structured data from a resume$0.0049$0.056
Debug a stack trace with context$0.0077$0.084
Per-call cost using published token counts for each task. Real-world prompts vary.