open god mode ▮

Inference provider ranking

# providers

78 providers serving 104 open-weights models, aggregated from live per-model offers. verified = models the provider's own catalog/price API confirms · top-10 = how many of the current overall top 10 they host · cheapest on = models where they have the best blended price · vs best $ = median premium over each model's cheapest offer (1.00× = usually the cheapest). Click a column to sort. Machine-readable: providers.json.

# provider hq models verified top-10 cheapest on vs best $ ctx (med) tps (med) ttft (med) uptime
1 Novita US 69 67 10 13 1.39× 200k 37 1132ms 99.7%
2 DeepInfra US 68 49 10 26 1.10× 198k 57 285ms 99.5%
3 SiliconFlow SG 46 42 10 4 1.38× 256k — — 99.6%
4 Featherless US 45 — 7 — — — — — —
5 Venice US 36 33 9 3 1.62× 250k — — 99.3%
6 Together US 32 — 8 1 1.71× 256k — — 99.0%
7 Alibaba SG 31 — 6 5 1.71× 256k — — 99.9%
8 AtlasCloud US 29 29 10 1 1.33× 256k — — 99.6%
9 Amazon Bedrock US 29 26 1 6 1.36× 128k — — 99.8%
10 Nebius NL 27 13 7 3 1.25× 125k — — 91.3%
11 Parasail US 24 — 6 1 1.81× 256k — — 99.8%
12 Phala US 23 23 7 1 1.71× 256k — — 96.5%
13 Wandb — 20 — 6 3 1.91× 256k — — —
14 StreamLake CN 19 — 8 3 1.27× 250k — — 99.4%
15 BaseTen US 17 — 7 — 1.47× 262k 166 273ms 98.5%
16 Azure US 17 1 4 3 3.20× 144k — — 73.1%
17 DigitalOcean — 16 — 7 1 1.69× 253k — — 99.4%
18 GMICloud US 15 — 8 5 1.33× 256k — — 98.6%
19 Fireworks US 14 — 8 — 2.05× 378k 115 512ms 98.7%
20 Google US 14 — — — 2.47× 231k — — 99.8%
21 Cloudflare US 12 — 3 1 4.05× 131k — — 99.8%
22 Scaleway — 12 — 1 1 3.76× 164k 92 426ms —
23 CoreWeave US 11 — 4 2 1.94× 256k — — 99.9%
24 Crusoe US 11 — 3 1 2.67× 256k — — 99.4%
25 Hyperbolic — 11 4 — 3 2.08× 128k — — —
26 Qwen Ai Platform — 10 — 5 — 1.72× 240k — — —
27 Qwencloud — 10 — 5 — 1.72× 240k — — —
28 Chutes US 10 8 4 — 1.97× 256k — — 95.6%
29 SambaNova US 10 6 1 — 3.92× 128k — — 98.2%
30 Nscale — 10 — — 5 1.01× 84k 60 653ms —
31 OVHcloud — 9 7 — 1 2.20× 128k 45 308ms —
32 Groq US 9 3 — — 2.98× 128k — — 99.7%
33 Baidu CN 8 — 6 2 1.16× 640k — — 100.0%
34 Vultr — 8 — 3 3 1.71× 256k — — —
35 Z.AI SG 8 — 3 2 1.36× 198k — — 99.6%
36 Mistral FR 7 1 1 4 1.00× 256k — — 100.0%
37 Replicate — 7 — — 1 3.94× 64k — — —
38 DeepSeek CN 6 — 2 2 1.45× 581k — — 100.0%
39 Tensormesh — 6 — 2 — 2.47× 80k — — —
40 Minimax SG 6 — 1 2 1.26× 200k — — 99.7%
41 Moonshot AI SG 5 — 3 — 1.33× 256k — — 100.0%
42 Friendli US 5 — 2 — 1.54× 198k — — 99.9%
43 NextBit ES 5 — 2 — 2.02× 1024k — — 99.6%
44 Libertai — 5 — 1 2 1.12× 256k — — —
45 AkashML — 5 — — 2 1.41× 128k — — 99.7%
46 Tencent CN 4 — 3 2 1.14× 1000k — — 100.0%
47 Inceptron SE 4 — 3 — 2.55× 640k — — 99.4%
48 Nvidia US 4 — 2 1 1.44× 640k — — —
49 Mancer 2 — 4 — 1 — 2.71× 576k — — 97.8%
50 Darkbloom — 4 — — 4 1.00× 192k — — 99.9%
51 Mara US 4 — — 1 3.34× 128k — — 95.1%
52 ModelRun US 3 — 2 — 3.18× 256k — — 100.0%
53 Gmi — 3 — — 1 1.32× 192k — — —
54 Decart US 2 — 2 — 1.95× 640k — — 99.9%
55 Makora — 2 — 1 1 1.43× 1024k — — 97.1%
56 Modal US 2 — 1 1 1.09× 1024k — — 99.9%
57 Morph US 2 — 1 — 1.97× 1024k — — 97.2%
58 Reka — 2 — 1 — 3.65× 256k — — 96.8%
59 Sail Research US 2 — 1 — 1.26× 1024k — — 99.8%
60 Wafer US 2 — 1 — 1.63× 1024k — — 99.4%
61 Nex AGI CN 2 — — 2 1.00× 256k — — 98.8%
62 Google Vertex — 2 — — 1 2.29× 145k — — —
63 Bedrock Mantle — 2 — — — 3.66× 128k — — —
64 Cerebras US 2 1 — — 7.12× 128k — — 100.0%
65 Xiaomi CN 2 — — — 1.30× 1024k — — 97.2%
66 Ambient — 1 — 1 — 1.80× 198k — — 96.1%
67 Doubleword — 1 — 1 — 4.08× 1024k — — —
68 Ionstream US 1 — 1 — 2.60× 1024k — — 99.9%
69 Scx — 1 — 1 — 1.81× 1024k — — —
70 Arcee AI US 1 — — 1 1.00× 256k — — 100.0%
71 Lambda — 1 — — 1 1.00× 128k — — —
72 LongCat — 1 — — 1 1.00× 1000k — — —
73 Poolside — 1 — — 1 1.00× 1024k — — 100.0%
74 IBM watsonx — 1 — — — 4.28× 128k — — —
75 Io Net US 1 — — — 2.07× 256k — — 100.0%
76 OpenInference US 1 — — — 1.24× 1024k — — 99.9%
77 Relace — 1 — — — 1.50× 1024k — — 98.9%
78 Snowflake — 1 — — — 2.46× 125k — — —

default order: model coverage, then top-10 coverage, then price wins. Prices are blended $/M tokens (3:1 input:output) from each provider's live offers; a model's per-provider offers are on its own page.