Crusoe vs DeepInfra: LLM provider comparison

At a glance

Metric winners

There is no overall score. Each winner answers one specific question using only directly comparable data.

Fastest on shared models

CCrusoe

1.90× typical advantage

6 exact models with complete recent speed data

Lowest token cost

DDeepInfra

$0.6733 / 1M

7 exact models using a 1K-input/500-output mix

Most models available

DDeepInfra

85 models

Complete catalog coverage across all indexed modalities

Shared-model benchmark summary

Metric	CCrusoe☆☆☆☆☆(0)	DDeepInfra☆☆☆☆☆(0)
Speed	Crusoe15.82 s	DeepInfra27.30 s
TTFT	Crusoe0.55 s	DeepInfra0.78 s
TPS	Crusoe33.0 tok/s	DeepInfra19.0 tok/s
Uptime	Crusoe99.55%	DeepInfra98.47%
Blended	Crusoe$0.8414 / 1M	DeepInfra$0.6733 / 1M

Model	CCrusoe☆☆☆☆☆(0)	DDeepInfra☆☆☆☆☆(0)
M Kimi K2.6Winner · DeepInfra	Crusoe Speed— Worst comparable value 34.64 s TTFT— Worst comparable value 1.31 s TPS— Worst comparable value 15.0 tok/s Uptime— Worst comparable value 56.00% Context 262.1K Route crusoe/bf16 Blended— Best comparable value $1.6333 / 1M	DeepInfra Speed— Best comparable value 14.24 s TTFT— Best comparable value 0.73 s TPS— Best comparable value 37.0 tok/s Uptime— Best comparable value 100.00% Context 262.1K Route deepinfra/fp4 Blended— Worst comparable value $1.6667 / 1M
ZA GLM 5.1Winner · Crusoe	Crusoe Speed— Best comparable value 5.89 s TTFT— Best comparable value 0.39 s TPS— Best comparable value 91.0 tok/s Uptime— Best comparable value 100.00% Context 202.8K Route crusoe/fp8 Blended— Worst comparable value $2.2667 / 1M	DeepInfra Speed— Worst comparable value 12.79 s TTFT— Worst comparable value 0.89 s TPS— Worst comparable value 42.0 tok/s Uptime— Worst comparable value 99.68% Context 202.8K Route deepinfra/fp4 Blended— Best comparable value $1.8667 / 1M
G Gemma 4 31BWinner · DeepInfra	Crusoe Speed— Worst comparable value 84.17 s TTFT— Worst comparable value 0.83 s TPS— Worst comparable value 6.0 tok/s Uptime— Worst comparable value 91.12% Context 262.1K Route crusoe Blended— Worst comparable value $0.2267 / 1M	DeepInfra Speed— Best comparable value 42.50 s TTFT— Best comparable value 0.83 s TPS— Best comparable value 12.0 tok/s Uptime— Best comparable value 97.49% Context 262.1K Route deepinfra/turbo Blended— Best comparable value $0.2033 / 1M
N Nemotron 3 Nano 30B A3B	Crusoe Speed N/a TTFT N/a TPS N/a Uptime— Worst comparable value 0.00% Context 262.1K Route crusoe/fp8 Blended $0.1 / 1M	DeepInfra Speed 8.02 s TTFT 1.17 s TPS 73.0 tok/s Uptime— Best comparable value 99.23% Context 262.1K Route deepinfra/fp4 Blended $0.1 / 1M
Q Qwen3 235B A22B Instruct 2507Winner · Crusoe	Crusoe Speed— Best comparable value 11.96 s TTFT— Best comparable value 0.33 s TPS— Best comparable value 43.0 tok/s Uptime— Best comparable value 99.58% Context 262.1K Route crusoe/bf16 Blended— Worst comparable value $0.4133 / 1M	DeepInfra Speed— Worst comparable value 25.36 s TTFT— Worst comparable value 0.36 s TPS— Worst comparable value 20.0 tok/s Uptime— Worst comparable value 98.47% Context 262.1K Route deepinfra/fp8 Blended— Best comparable value $0.2433 / 1M
D DeepSeek V3 0324Winner · DeepInfra	Crusoe Speed— Best comparable value 17.05 s TTFT— Best comparable value 0.39 s TPS— Best comparable value 30.0 tok/s Uptime— Best comparable value 99.55% Context 163.8K Route crusoe/bf16 Blended— Worst comparable value $0.8333 / 1M	DeepInfra Speed— Worst comparable value 29.24 s TTFT— Worst comparable value 1.46 s TPS— Worst comparable value 18.0 tok/s Uptime— Worst comparable value 43.42% Context 163.8K Route deepinfra/fp4 Blended— Best comparable value $0.46 / 1M
ML Llama 3.3 70B InstructWinner · Crusoe	Crusoe Speed— Best comparable value 14.59 s TTFT— Worst comparable value 0.70 s TPS— Best comparable value 36.0 tok/s Uptime— Best comparable value 99.91% Context 131.1K Route crusoe/bf16 Blended— Worst comparable value $0.4167 / 1M	DeepInfra Speed— Worst comparable value 63.18 s TTFT— Best comparable value 0.68 s TPS— Worst comparable value 8.0 tok/s Uptime— Worst comparable value 97.29% Context 131.1K Route deepinfra/turbo Blended— Best comparable value $0.1733 / 1M

A per-model winner combines blended price and estimated 500-token response time with equal proportional weight. Ties and rows missing either measurement receive no badge. Comparison data calculated Jul 24, 2026, 6:19 PM UTC. Values use one deterministic route per provider and model; missing measurements remain visible as N/a.