Providers on Liquid Inference

Each provider posts standing offers on Liquid Inference. For each request, the exchange ranks the offers that meet the buyer's rules by estimated charge, and the buyer pays the lowest.

The exchange probes each deployment, compares the answers of each offer with a baseline for its model, and runs benchmarks. Each provider page states the results.

Providers

• Architect: 754 models, 765 offers, median output rate $1.197 per million tokens, regions us, declared precisions provider-managed. 7 of 765 offers match the fingerprint of the model. 0 have a finished evaluation. Data: Zero retention, checked by the exchange; No training, checked by the exchange.
• EmpirioLabs AI: 103 models, 132 offers, median output rate $2.400 per million tokens, regions global, declared precisions bf16, fp16, fp4, fp8, int2, int8, mixed, provider-managed. 3 of 132 offers match the fingerprint of the model. 0 have a finished evaluation. Data: Zero retention, declared by the provider; No training, declared by the provider; GDPR DPA, declared by the provider; CSA STAR, declared by the provider.
• Pera: 35 models, 36 offers, median output rate $0.411 per million tokens, regions asia, us, declared precisions bf16, fp4, fp8, mxfp4, provider-managed. 14 of 36 offers match the fingerprint of the model. 1 has a finished evaluation. Data: Zero retention, declared by the provider; No training, declared by the provider.
• Greenference: 10 models, 10 offers, median output rate $0.328 per million tokens, regions eu-central, declared precisions fp8. 0 of 10 offers match the fingerprint of the model. 0 have a finished evaluation. Data: Zero retention, declared by the provider; No training, declared by the provider.
• Munito AI: 10 models, 10 offers, median output rate $0.273 per million tokens, regions global, declared precisions fp8, nvfp4. 6 of 10 offers match the fingerprint of the model. 0 have a finished evaluation. Data: Zero retention, declared by the provider; No training, declared by the provider.
• engy: 9 models, 9 offers, median output rate $0.140 per million tokens, regions asia, declared precisions fp4, fp8. 3 of 9 offers match the fingerprint of the model. 0 have a finished evaluation.
• SOKKAN: 8 models, 8 offers, median output rate $0.809 per million tokens, regions eu, declared precisions fp8, provider-managed. 2 of 8 offers match the fingerprint of the model. 0 have a finished evaluation. Data: Zero retention, declared by the provider; No training, declared by the provider; GDPR DPA, declared by the provider.
• akash: 7 models, 7 offers, median output rate $0.900 per million tokens, regions us-west, declared precisions bf16, fp4, fp8. 3 of 7 offers match the fingerprint of the model. 0 have a finished evaluation.
• RelayGPU: 7 models, 7 offers, median output rate $5.148 per million tokens, regions global, declared precisions provider-managed. 0 of 7 offers match the fingerprint of the model. 0 have a finished evaluation. Data: Kept 90 days, declared by the provider; No training, declared by the provider.
• TensorX: 7 models, 7 offers, median output rate $2.400 per million tokens, regions eu, declared precisions fp8, mxfp4, mxfp8, nvfp4. 1 of 7 offers match the fingerprint of the model. 0 have a finished evaluation. Data: Zero retention, declared by the provider; No training, declared by the provider.
• MarsCompute: 6 models, 6 offers, median output rate $2.100 per million tokens, regions global, us, declared precisions fp8, provider-managed. 1 of 6 offers match the fingerprint of the model. 0 have a finished evaluation. Data: Zero retention, declared by the provider; No training, declared by the provider.
• blackfuel: 5 models, 8 offers, median output rate $0.290 per million tokens, regions eu, global, declared precisions bf16, fp8, nvfp4. 3 of 8 offers match the fingerprint of the model. 0 have a finished evaluation. Data: Zero retention, declared by the provider; No training, declared by the provider; SOC 2 Type II, declared by the provider; ISO 27001, declared by the provider.
• Boundless: 4 models, 4 offers, median output rate $0.201 per million tokens, regions us, declared precisions fp8. 3 of 4 offers match the fingerprint of the model. 0 have a finished evaluation. Data: Zero retention, declared by the provider; No training, declared by the provider.
• deeprelay: 4 models, 4 offers, median output rate $1.665 per million tokens, regions ap-southeast, cn, declared precisions provider-managed. 0 of 4 offers match the fingerprint of the model. 0 have a finished evaluation. Data: Zero retention, declared by the provider; No training, declared by the provider.
• zro: 4 models, 5 offers, median output rate $1.500 per million tokens, regions eu, us, declared precisions fp8, provider-managed. 2 of 5 offers match the fingerprint of the model. 1 has a finished evaluation. Data: No training, declared by the provider.
• amazee.ai: 3 models, 3 offers, median output rate $0.380 per million tokens, regions us, declared precisions nvfp4. 1 of 3 offers match the fingerprint of the model. 0 have a finished evaluation. Data: Zero retention, declared by the provider; No training, declared by the provider; SOC 2 Type II, declared by the provider; ISO 27001, declared by the provider.
• Alpha.sh: 2 models, 2 offers, median output rate $1.775 per million tokens, regions us-west, declared precisions fp8. 0 of 2 offers match the fingerprint of the model. 0 have a finished evaluation.
• Fortytwo: 2 models, 2 offers, median output rate $0.400 per million tokens, regions us, declared precisions custom. 1 of 2 offers match the fingerprint of the model. 0 have a finished evaluation. Data: Zero retention, declared by the provider; No training, declared by the provider.
• Valar: 2 models, 2 offers, median output rate $1.339 per million tokens, regions us-east, declared precisions fp8. 1 of 2 offers match the fingerprint of the model. 0 have a finished evaluation. Data: Zero retention, declared by the provider; No training, declared by the provider; SOC 2 Type II, declared by the provider; GDPR DPA, declared by the provider; PCI DSS, declared by the provider.
• Butler: 1 model, 1 offer, regions us-east, declared precisions nvfp4. 0 of 1 offer matches the fingerprint of the model. 0 have a finished evaluation. Data: Zero retention, declared by the provider; No training, declared by the provider.
• DreamcatcherAI: 1 model, 1 offer, median output rate $8.899 per million tokens, regions us-central, declared precisions BFP8. 0 of 1 offer matches the fingerprint of the model. 0 have a finished evaluation. Data: Kept 30 days, declared by the provider; No training, declared by the provider.
• ElastixAI: 1 model, 1 offer, median output rate $0.250 per million tokens, regions us-west, declared precisions int4. 0 of 1 offer matches the fingerprint of the model. 0 have a finished evaluation. Data: Zero retention, declared by the provider; No training, declared by the provider.
• instant: 1 model, 1 offer, median output rate $0.333 per million tokens, regions us, declared precisions fp8. 0 of 1 offer matches the fingerprint of the model. 0 have a finished evaluation.
• muna: 1 model, 1 offer, median output rate $0.300 per million tokens, regions us, declared precisions fp4. 0 of 1 offer matches the fingerprint of the model. 0 have a finished evaluation. Data: Zero retention, declared by the provider; No training, declared by the provider.
• Silicon Exchange: 1 model, 1 offer, median output rate $9.700 per million tokens, regions us-east, declared precisions provider-managed. 0 of 1 offer matches the fingerprint of the model. 0 have a finished evaluation. Data: Zero retention, declared by the provider; No training, declared by the provider.
• umans-ai: 1 model, 1 offer, median output rate $0.600 per million tokens, regions uk, declared precisions fp8. 1 of 1 offer matches the fingerprint of the model. 0 have a finished evaluation.

Related pages

• All models
• Discover the best model for your budget
• Quickstart
• llms.txt: the index of every page, as Markdown
• llms-full.txt: every page in one file
• This page as Markdown