comparing 3 of 127

Pick the model that fits the job.

anthropic / opus

Claude 4.5 Opus

Best-in-class reasoning. The thinker.

Use this model →

specs

ctx: 1M
out: 64K
lat: 2.4s
pricing: 0.04 KM /1K

benchmarks

reason

96

code

92

vision

84

speed

62

best for

avoid if

openai / flagship

GPT-5

Generalist. Reliable. Plays well with tools.

Use this model →

specs

ctx: 400K
out: 32K
lat: 3.1s
pricing: 0.07 KM /1K

benchmarks

reason

92

code

95

vision

91

speed

58

best for

avoid if

google / pro

Gemini 2.5 Pro

Massive context, great with your data.

Use this model →

specs

ctx: 2M
out: 32K
lat: 1.8s
pricing: 0.03 KM /1K

benchmarks

reason

88

code

86

vision

95

speed

84

best for

avoid if