Integrations
Every provider in the path. Every bill in sync.
16 native integrations across model providers, cloud platforms, billing engines, observability and agent frameworks. Model traffic needs one base URL change; everything else is configured in the console.
OpenAI1.2M req todayconnected
Anthropic860k req todayconnected
AWS Bedrock212k req todayconnected
Stripe Billing48k meter eventsconnected
Datadogmetrics + eventsconnected
16 of 16 integrations
Google Gemini and Vertex AIModel providerGovern Gemini traffic from the Gemini API or Vertex AI with one base URL change. PALM SOUQ meters cached content tokens and thinking tokens, and handles the long-context price bands Gemini models use.
Azure OpenAICloud platformKeep your Azure OpenAI deployments, regions and enterprise agreement. PALM SOUQ sits in front of each deployment, attributes usage per tenant and enforces quotas before your shared Azure TPM allocation runs dry.
DeepSeekModel providerDeepSeek's OpenAI-compatible API drops into PALM SOUQ with a base URL change. The gateway meters disk cache hits and misses and surfaces reasoner thinking output so low prices stay low per tenant.
GroqModel providerGroq's fast inference for Llama and other open-weight models runs through PALM SOUQ's OpenAI-compatible gateway. Tenant attribution and velocity limits matter more when tokens arrive this quickly.
Mistral AIModel providerMeter Mistral Large, Medium, Small and Codestral per tenant through PALM SOUQ. Works with the official mistralai SDK or any OpenAI-compatible client, with EU data residency options at the gateway.
Together AIModel providerRun open-weight models on Together AI and govern them per tenant. PALM SOUQ's OpenAI-compatible route meters serverless inference and lets you mix Together models into fallback chains.
MetronomeBillingStream per-tenant usage events into Metronome's ingest API with the properties your billable metrics filter on: model, feature, token type and cost. Contracts, commits and credits stay in Metronome.
OrbBillingSend PALM SOUQ usage to Orb's event ingestion with idempotency keys and rich properties, then price tokens, cached tokens and premium models with Orb's billable metrics.
LagoBillingUse Lago, cloud or self-hosted, as the billing engine for token usage. PALM SOUQ posts events to your Lago instance with billable metric codes and subscription ids already resolved.
OpenTelemetryObservabilityExport PALM SOUQ spans and metrics over OTLP to any OpenTelemetry backend, using the GenAI semantic conventions for model, token usage and operation names, plus PALM SOUQ attributes for tenant and cost.
LangChain and LangGraphAgent frameworkPut PALM SOUQ under LangChain chat models and LangGraph agents by setting the base URL on the model class. Loop heuristics then watch every step of the graph, not just the final answer.
In the request path
Model providers and clouds. Your SDK points at gateway.palmsouqltd.com; breakers, quotas and fallbacks act before the provider is called. Under 5 ms added.
Off the request path
Billing engines and observability. Usage and decisions are exported asynchronously from the region that metered them, so a slow downstream never slows a model call.
In your code
Agent frameworks. LangChain and LangGraph models take a base URL and headers, so every node is governed and tagged by feature.
Custom providers
Self-hosted or niche model? Register it.
Any endpoint that speaks the OpenAI chat completions format can sit behind PALM SOUQ: vLLM, TGI, an internal fine-tune, or a regional host. Give it a price per million tokens and it is metered, governed and usable in fallback chains like any native provider.
# palmsouq.yaml (custom OpenAI-compatible provider)
providers:
- id: vllm-eu
kind: openai_compatible
base_url: https://llm.internal.example.eu/v1
auth: { header: Authorization, secret_ref: vault://vllm/token }
pricing: # your own cost per 1M tokens
llama-3.3-70b: { input: 0.42, output: 0.42 }
region: eu-centralConnect your first provider in two minutes.
Route your agent traffic through PALM SOUQ in 2 minutes. Change one base URL, pass a tenant header, and every token is attributed and governed.
14-day free production trial · 100k requests included · no card required