Compare

Comparison table

All 20 indexed providers, side by side. Use the filters to narrow by category, OpenAI-compatibility or compliance certification, and click any column header to re-sort. Figures follow the confidence legend on themethodology page.

Ranked by composite reputation index —see Methodology. Click any header to re-sort.

Comparison of 20 AI inference providers.

OpenAI/openai	Frontier lab	$6.25	480	90	99.8	✓	91
Anthropic/anthropic	Frontier lab	$9.00	520	78	99.8	·	90
OpenRouter/openrouter	Router	$0.90	400	120	99.6	✓	88
Groq/groq	Specialist hardware	$0.69	120	480	99.6	✓	87
Together AI/together-ai	Aggregator	$0.88	340	130	99.7	✓	86
Fireworks AI/fireworks-ai	Aggregator	$0.90	300	145	99.7	✓	85
Modal/modal	GPU cloud	—	700	100	99.7	·	85
Google Vertex AI/google-vertex	Hyperscaler	$3.13	610	85	99.9	✓	84
Baseten/baseten	GPU cloud	$0.80	280	140	99.6	✓	84
Mistral La Plateforme/mistral	Frontier lab	$4.00	430	95	99.6	✓	83
cheapestinference/cheapestinference	Subscription (flat-rate)	$0.30	340	120	99.5	✓	83
Cerebras Inference/cerebras	Specialist hardware	$0.70	110	520	99.5	✓	82
Perplexity API/perplexity	Specialist app	$1.00	650	80	99.5	✓	82
DeepInfra/deepinfra	Aggregator	$0.38	360	110	99.5	✓	81
Replicate/replicate	Aggregator	$1.70	900	60	99.3	·	80
SambaNova Cloud/sambanova	Specialist hardware	$0.90	130	400	99.4	✓	79
Nebius AI Studio/nebius	GPU cloud	$0.50	370	118	99.5	✓	79
Lepton AI/lepton-ai	GPU cloud	$0.50	330	125	99.4	✓	78
Novita AI/novita-ai	Aggregator	$0.35	380	105	99.4	✓	77
Hyperbolic/hyperbolic	GPU cloud	$0.40	350	115	99.3	✓	76

Reading this table

Blended $/1M — mean of input & output price for the reference model; — = billed by GPU-time.
TTFT ms — reference time-to-first-token; lower is faster.
tok/s — reference sustained output throughput; higher is faster.
Reputation — editorial composite index 0–100 with 90-day trend (▲/■/▼), not a first-party metric.