Best LLM Inference Providers

Vendor profileChecked Nov 8, 2026

Baseten

Baseten Labs, Inc. · baseten.co · Last checked:

  • Serverless per-token API (Model APIs)
  • Dedicated GPU deployments
  • Bring-your-own-model (Truss)

Company status

Active, privately held.

Sources 1, 2, 3, 4, 5 · checked 2026-11-08

Capabilities

  • Model APIs for open-weight models with OpenAI Chat Completions and Responses compatibility (Anthropic Messages in beta)
  • Dedicated inference on per-minute GPUs (T4 through B200) with autoscaling and scale to zero
  • Truss packaging to deploy your own or fine-tuned models
  • Zero data retention by default for synchronous inference; async inputs stored up to 72 hours
  • Region selection for GPU deployments with regional endpoints for data residency; self-hosted and single-tenant options

Sources 1, 2, 3, 4, 5 · checked 2026-11-08

Who it tends to fit

  • Teams that want both a per-token API and dedicated GPUs for custom models in one vendor
  • Buyers with data-residency needs who want deployments pinned to a region

Sources 1, 2, 3, 4, 5 · checked 2026-11-08

Pricing

Published per-token pricing for Model APIs and per-minute GPU pricing on https://www.baseten.co/pricing/ (for example, H100 80 GiB listed at $0.10833 per minute for dedicated inference). Volume and reserved capacity by quote.

Sources 1, 2, 3, 4, 5 · checked 2026-11-08

Published certifications

Listed as stated by the vendor. Scope can differ by product or subsidiary; request current reports.

Sources 1, 2, 3, 4, 5 · checked 2026-11-08

Limitations to check

  • Async inference inputs are stored until processed (up to 72 hours)
  • Custom regions, SLAs, and reserved capacity require a sales agreement
  • Model API catalog changes often; confirm the model you need is listed

Sources 1, 2, 3, 4, 5 · checked 2026-11-08

Sources

  1. Pricing page: Model APIs priced per 1M tokens; dedicated GPU pricing per minute (H100 $0.10833/min); custom regions and SLAs by agreementwww.baseten.co/pricing/Checked 2026-11-08
  2. Security page: zero data retention by default for synchronous inference; async inputs up to 72 hours; region selection; Baseten Cloud, single tenant, self hosted; SOC 2 Type II, SOC 3, HIPAA, GDPR, PCI DSS, CCPAwww.baseten.co/security/Checked 2026-11-08
  3. Trust Center lists SOC 2, ISO 27001:2022, HIPAA, GDPR, CCPA, PCI DSS SAQ D, with SOC 2 Type 2 and ISO 27001 documentstrust.baseten.coChecked 2026-11-08
  4. Model APIs support OpenAI Chat Completions, OpenAI Responses, and Anthropic Messages (beta)docs.baseten.co/development/model-apis/overviewChecked 2026-11-08
  5. Truss used to package and deploy your own models on dedicated infrastructuredocs.baseten.co/development/model/overviewChecked 2026-11-08

Report a correction: [email protected]. Facts are re-checked at least every 90 days.