Partners

Build with us, on us, or alongside us.

Three programs for three different shapes of partnership. All real, none of them require a 12-week procurement process.

01 · Build

Application partners.

You ship a product on top of Tomoul. We make sure the substrate doesn't get in your way and your users get to a working call quickly.

  • +$5,000 in launch credits, refilled on traction
  • +Dedicated Slack channel with on-call engineer
  • +Co-marketing on launch — case study, blog, talk
  • +Pre-release access to new models and regions
02 · Distribute

Model authors.

You publish open weights. We host them well, route demand to them, and share revenue when the model is in our menu.

  • +Hand-tuned hosting on dedicated SKUs
  • +50 / 50 net-revenue share above hosting cost
  • +Co-listed on the model catalog with your bio and links
  • +Eval suite and benchmarks published, signed
03 · Infrastructure

Infra partners.

GPU clouds, payment rails, observability tools, IDE plugins, agentic frameworks. Distribution is mutual; integrations are real.

  • +Reference architecture and joint engineering
  • +Listing on the integrations page with one-click setup
  • +Joint customer support for shared accounts
  • +Field events — KubeCon, Africa AI Week, AI Engineer

Who we ship with today.

A short, real list. We do not put logos on this page until the integration works.

cloudflare
vercel
supabase
linear
railway
upstash
langchain
llama-index
openrouter
continue
cline
aider

Cases.

Three early customers, in their own words. Long-form case studies linked on each row.

vox.africa
Lagos · Media · 4M MAU
“We replaced four pipelines with one Tomoul endpoint. Yoruba TTS that doesn't sound like a robot is the feature our users actually mention in app store reviews.”
— Chiamaka N., Head of Engineering · Read the case study →
$84k
Annual GPU savings
4.2×
TTS throughput
kweli.health
Nairobi · Health · Telemedicine
“Swahili embeddings that actually retrieve. We tested four vendors; Tomoul was the only one whose bge-m3 variant was tuned for the languages our patients speak.”
— David O., CTO · Read the case study →
31%
Retrieval recall lift
EU
Residency for partners
tilbake.no
Oslo · Devtools · Coding agents
“Our coding agent makes 60 inference calls per task. p99 TTFT was killing us on the old provider. Tomoul's tail at fra1 is the cleanest we have measured.”
— Sigrid Halvorsen, Founder · Read the case study →
184ms
p99 TTFT · chat
99.97%
Uptime · 90 days
Apply

Tell us what you'd want to build.

One paragraph is enough. Mention the program, the product, the shape of the partnership, and a link. We respond within five business days.