Compare

Gemini 2.5 Ultra vs Claude Haiku 4.5

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

Verdict

Gemini 2.5 Ultra leads on LMSYS ELO (1385 vs 1305). Claude Haiku 4.5 is ~6.6× cheaper on a 3:1 input:output blend.

Google DeepMind

2M-token context, native video understanding, and Google’s deepest multimodal stack. The long-context king.

2M context windowNative videoBest multimodal reasoning

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$7.00$0.80

Output / 1M$21.00$4.00

Context2.0M200K

Max output64K16K

LicenseProprietaryProprietary

Released2026-02-052025-10-01

Benchmarks

LMSYS ELO1385.01305.0

MMLU Pro90.481.4

HumanEval90.287.0

SWE-Bench58.349.2

MATH92.076.8

MMMU82.1—

IFEval89.787.3

Per-task cost

Summarize a 1-hour meeting transcript$0.115$0.014

Review a 500-line pull request$0.098$0.014

Answer a customer support ticket$0.038$0.0056

Extract structured data from a resume$0.056$0.0080

Debug a stack trace with context$0.084$0.013

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

Add up to four models, tweak tasks live.