Meldport and OpenRouter: compare the workload
Start with the model and protocol your application needs, then compare billable usage, funding costs and operational controls.
This page is published by Meldport about its own service. It is not an independent review or an endorsement by OpenRouter. Vendor feature and credit-fee information was checked on 2 November 2026. The listed-price dataset was checked 2026-11-02T18:09:39.794Z.
What to verify before moving a request
- Endpoint and model: send a bounded test to
POST /v1/chat/completionsusing the exact destination model ID from the Meldport catalog. OpenRouter IDs and protocol options can differ. - Streaming and tools: confirm the client reads the full event sequence, then test one tool call and its matching result on the selected model.
- Billing: reconcile one test request against Usage and check cached rates, per-request minimums and wallet fallback separately from funding fees.
Use the migration checklist before switching production traffic.
Same hypothetical usage, separate billing assumptions
The example uses 10,000 ordinary input tokens and 1,000 output tokens per request, 100 requests per day for 30 days: 3,000 requests in total. It assumes no cache hits, batch processing, media or retries. It is a reproducible illustration, not a measurement of customer spending or model quality.
A fresh comparable price dataset is not available. Workload totals are withheld until both rate evidence and model limits can be verified. Inspect current Meldport rates and check OpenRouter's live model page.
Matches use exact legacy ID suffixes; ambiguous aliases and batch variants are excluded. Matching an ID does not certify an identical upstream provider, model build, region, context policy, uptime or output quality. A model that cannot fit this workload, lacks a published minimum or has expired evidence is excluded. Only a subset of eligible rows is shown; missing rows are not wins for either service. Cached input is unavailable in this cross-vendor snapshot and is not silently treated as free.
Fees, free usage and protocols
| Decision | Meldport | OpenRouter |
|---|---|---|
| Token rates | Published per-model input, output, cache and minimum rates. The calculator shows its assumptions. | Published model/provider rates. Provider routing and model availability can differ from Meldport's routes. |
| Funding costs | Payment rail, currency conversion and any fee are shown in the wallet checkout quote before payment. They are excluded from the token estimate above. | The pricing page checked on 2 November lists a 5.5% Standard credit-purchase fee and an 8% Business fee. These are separate from inference prices; confirm the current plan and checkout terms. |
| Free models | One shared monthly token pool, verified-email eligibility and optional account/key wallet fallback. See the live free policy and aliases. | Free variants have their own request-count limits and model availability. Check OpenRouter's free-model rules; they are not a Meldport-style monthly token pool. |
| Application compatibility | The documented LLM protocol is Chat Completions. Native Anthropic Messages or Responses clients need separate protocol support; changing a base URL alone is insufficient. | Check the protocol, client integration and routing features documented for the exact model and endpoint you intend to use. |
| Media workflows | Image, video and audio generation use the documented configuration, quote and job contracts. | Check the specific model and modality contract. This token-price table does not compare media pricing or outputs. |
When each service can fit
Meldport can fit a workload whose required model is published here, whose client supports the documented API and whose measured cost and limits suit your use. OpenRouter can fit a model, provider-routing option, protocol or client integration outside Meldport's current documented inventory. A lower rate for one token mix does not establish a universal cheapest gateway.
Test your own prompts, tool round trips, context use and error handling before moving a production workload. Keep the original provider available until request results and billing reconcile. The migration guide covers the transfer procedure; this page covers the purchasing decision.