Koyeb vs Together GPU Clusters

Koyeb 31

No downloadable OpenAPI file. The API reference at koyeb.com/docs/reference/api renders only by script, and the generated docs in the Go and Python client repositories aren't a published contract (0, was 10). No llms.txt (www.koyeb.com/llms.txt and /docs/llms.txt both 404) and no Markdown twins (0). Guides on scale to zero and autoscaling say when to use each setting (10, held). Typed parameters with required bodies in the Go client's docs, but service types and statuses are free strings with no listed values (8, held). CLI examples throughout; no documented error responses found (5, held). /v1 paths, but the public changelog stopped on 27 February 2026 (8, held). 0 + 0 + 10 + 8 + 5 + 8 = 31.

Together GPU Clusters 92

Public OpenAPI 3.1 for /v1/compute/clusters and shared volumes, MIT, embedded in every reference page (25). llms.txt and a Markdown twin of every page (10). The docs say when to choose Kubernetes or Slurm and each capacity type, and the management page carries instructions for agents on reporting live availability and prices (18). Enums for cluster type, billing type, GPU type and status, required fields marked (13). CLI, Python, TypeScript and console steps on each task, with 402 and 403 cases documented, though no error catalogue for the clusters API (11). Versioned SDKs and a dated changelog with cluster entries (15). 25+10+18+13+11+15 = 92.

Koyeb 50

Personal access tokens and organisation tokens sent as Bearer; no scopes found (20). RBAC only on Enterprise; no read-only token (5). Runs your own containers and returns their output (10). Audit logs only on Enterprise (10). ISO 27001 and SOC 2 listed on the pricing page for Enterprise; www.koyeb.com/.well-known/security.txt is 404 and no disclosure policy, SECURITY.md or bug bounty found, so a bare certification claim (5). 20 + 5 + 10 + 10 + 5 = 50.

Together GPU Clusters 68

Project-scoped API keys, revocable with optional expiry, read as Together's other listings read the same key model (30). Admin and editor roles with no read-only role. Only admins get the admin kubeconfig, which has cluster-admin rights, and OIDC lets editors use kubectl under RBAC. Deletes run without confirmation when --non-interactive is passed (8). The API returns cluster state, not third-party content (10). Invoices itemise reserved, on-demand and storage charges, and no audit log for cluster operations was found (5). security.txt with a HackerOne route, per the 1 November check of the inference listing (15). 30+8+10+5+15 = 68.

Koyeb 50

Graded on the hosted lines for its platform API. Instatus page at status.koyeb.com with components (API, Control Panel, Build / Provisioning and one per region) and history back to 2022 in its feed (20). Since 7 August the feed holds nine incidents and one maintenance. The eight in September each name one region, one GPU type or some customers, or ran 6 to 52 minutes (8, 11, 15, 21, 22 and 23 September and two on 29 September). On 16 October Koyeb posted a separate incident, "The ongoing maintenance is causing API to fail with 500", on the API, Control Panel and Build components for 24 minutes, inside a maintenance window announced the day before for 07:00 to 08:00 UTC with "max 10m unavailability". The status page labels it an incident and the outage ran past the unavailability announced, so it counts as an incident, and the API failing with 500 is errors across the board, one major outage (10). No API rate limits found in the docs we could read (0). No 429 or backoff guidance found (0). The Scale plan lists a 99.9 per cent uptime SLA and Enterprise 99.99, with an SLA document at /docs/legal/sla setting service credits (10). Services and GPUs are generally available; light sleep is a preview (10). 20 + 10 + 0 + 0 + 10 + 10 = 50.

Together GPU Clusters 38

Read with the hosted lines. status.together.ai lists the website, the playground and serverless models with no component for GPU clusters, and a vendor page with no component for the product takes 15 in this run, as for Together's Code Sandbox and fine-tuning listings (15). No incident history for clusters can be read (5). No rate limits for the clusters API found (0). Together's docs give 429 and 503 backoff guidance for its APIs, per the inference listing's 1 November check, and the CLI has --non-interactive and --json for scripted retries, but creation has no idempotency key (8). No SLA for clusters found; the published 99% commitment covers provisioned inference throughput only (0). Generally available since 9 September 2025 and out of the CLI's beta paths since 22 October 2026, with preemptible nodes in preview (10). 15+5+0+8+0+10 = 38.