The Known Good Updated 29 Jul 2026

API providers

Who actually serves the models, what they charge for them, and how fast they answer. Prices and endpoint telemetry are ingested from OpenRouter — we do not run our own provider benchmarks.

73 hosts serving tracked models
All 73 of 73 providers
Provider Models served Mean output t/s Mean TTFT Lowest blended $/M
OpenAI · first party 71 63 3.5s $0.069
Novita 68 38 1.7s $0.000
DeepInfra 67 42 1.0s $0.022
Google · first party 54 83 2.0s $0.087
Alibaba · first party 46 62 0.9s $0.055
Azure · first party 46 52 4.7s $0.14
SiliconFlow 35 30 2.1s $0.075
Parasail 34 52 0.8s $0.060
Venice 30 48 1.8s $0.11
AtlasCloud 27 41 1.6s $0.17
Amazon Bedrock · first party 24 76 1.8s $0.061
Anthropic · first party 24 57 2.2s $1.00
Google AI Studio 23 121 3.2s $0.000
CoreWeave 20 75 0.6s $0.055
StreamLake 20 37 1.7s $0.084
Together 20 63 0.7s $0.075
Phala 19 44 1.6s $0.055
DigitalOcean 17 29 1.2s $0.14
Mistral · first party 17 54 0.4s $0.10
Cloudflare 14 50 0.7s $0.041
GMICloud 14 38 2.6s $0.100
Nebius 14 66 0.6s $0.10
NextBit 12 28 1.2s $0.060
Z.AI · first party 11 32 5.5s $0.42
Chutes 9 31 1.9s $0.18
Fireworks 9 87 1.0s $0.13
Groq 9 238 0.3s $0.058
Minimax · first party 8 46 1.2s $0.42
Morph 8 310 1.2s $0.17
Baidu · first party 7 44 1.1s $0.11
BaseTen 7 118 1.6s $0.20
Crusoe 7 54 0.4s $0.087
Nvidia · first party 7 68 1.0s $0.000
Friendli 6 65 0.7s $0.20
Io Net 6 45 2.6s $0.059
Mancer 2 6 24 0.9s $0.17
Poolside 6 39 1.4s $0.000
SambaNova 6 142 1.5s $0.34
AkashML 5 59 1.5s $0.12
Cohere · first party 5 42 0.7s $0.000
Ionstream 5 39 1.4s $0.17
Perplexity · first party 5 60 18.3s $1.00
xAI · first party 5 192 3.2s $1.25
AionLabs 4 52 1.0s $0.88
Inceptron 4 35 1.2s $0.34
Mara 4 121 2.2s $0.30
Moonshot AI · first party 4 43 2.3s $1.20
Seed · first party 4 59 1.1s $0.13
Ambient 3 38 4.0s $0.17
Cerebras 3 280 0.3s $0.45
Wafer 3 69 3.9s $1.50
Darkbloom 2 18 3.8s $0.000
Decart 2 76 0.5s $1.34
DeepSeek · first party 2 68 0.9s $0.17
Inflection 2 $4.38
ModelRun 2 60 0.6s $0.30
Nex AGI 2 104 1.0s $0.044
Reka · first party 2 16 48.8s $0.10
Relace 2 4 0.7s $0.95
Sail Research 2 26 1.1s $1.62
Xiaomi · first party 2 39 2.6s $0.17
AI21 · first party 1 20 0.7s $3.50
Arcee AI 1 27 0.7s $0.39
Inception 1 257 0.7s $0.38
Meta · first party 1 135 2.5s $2.00
OpenInference 1 15 3.1s $0.16
Perceptron 1 31 0.5s $0.49
Sakana AI 1 $11.25
StepFun 1 46 5.1s $0.44
Tencent · first party 1 54 2.3s $0.23
Upstage · first party 1 14 1.8s $0.26
Claude Platform on AWS 1 46 2.3s $10.00
Modal 1 41 2.3s $6.00
How to read this. A host appears here once we have seen it serve at least one tracked model. Mean output t/s and mean TTFT average that host's measurements across every model it serves, so a host that only serves small models will look faster than one carrying frontier models — compare hosts on a single model page for a like-for-like reading. Lowest blended $/M is the cheapest blended rate that host publishes for any model, at a 3:1 input:output blend. Figures are ingested from OpenRouter endpoint listings and telemetry, with metadata enriched from models.dev. See methodology and attribution.