Question 1

Does Fivo work with GPT-4o?

Accepted Answer

Yes. Full support for GPT-4o including chat completions, function calling, structured outputs, streaming, and vision. Save 5�20� on GPT-4o costs.

Question 2

Does Fivo work with GPT-4 Turbo?

Accepted Answer

Yes. Fivo supports GPT-4 Turbo with all features � 128K context, function calling, JSON mode, streaming. Same API, 5�20� cheaper.

Question 3

Does Fivo work with GPT-3.5 Turbo?

Accepted Answer

Yes. GPT-3.5 Turbo is fully supported. While already cheaper than GPT-4, Fivo still delivers 10-40x additional savings on GPT-3.5 queries.

Question 4

Does Fivo support OpenAI embeddings?

Accepted Answer

Yes. Fivo supports text-embedding-3-small, text-embedding-3-large, and text-embedding-ada-002. Embedding queries benefit from the same optimization.

Question 5

Does Fivo support OpenAI function calling?

Accepted Answer

Yes. Full support for function calling (tool use), parallel function calls, and structured outputs. Function definitions and response parsing work identically through Fivo.

Question 6

Does Fivo support OpenAI streaming?

Accepted Answer

Yes. Full SSE streaming support. Optimized streaming responses start in under 5ms � significantly faster time-to-first-token compared to direct OpenAI (200-800ms).

Question 7

Does Fivo support OpenAI JSON mode?

Accepted Answer

Yes. Structured outputs and JSON mode work identically through Fivo. Set response_format to json_object and Fivo handles it transparently.

Question 8

Does Fivo support OpenAI vision (image inputs)?

Accepted Answer

Yes. GPT-4o vision is supported. Send images in your messages and Fivo handles them. Image inputs are passed through to OpenAI when optimization is not applicable.

Question 9

Does Fivo work with fine-tuned OpenAI models?

Accepted Answer

Yes. Specify your fine-tuned model ID (ft: prefix) in requests as usual. Fivo optimizes fine-tuned model queries the same way as base models.

Question 10

How do I set up Fivo with the OpenAI Python SDK?

Accepted Answer

One line: client = OpenAI(base_url='https://api.Fivo.dev/v1', api_key='your-openai-key'). All other code stays exactly the same. Takes 30 seconds.

Question 11

How do I set up Fivo with the OpenAI Node.js SDK?

Accepted Answer

Set baseURL in the client: const openai = new OpenAI({ baseURL: 'https://api.Fivo.dev/v1', apiKey: 'your-key' }). All other code unchanged.

Question 12

Does Fivo support the OpenAI Assistants API?

Accepted Answer

Fivo currently optimizes the Chat Completions and Embeddings APIs. Assistants API support is on the roadmap. Most teams use Chat Completions directly for maximum control and Fivo compatibility.

Question 13

Does Fivo support OpenAI Batch API?

Accepted Answer

Fivo optimizes individual API calls. For OpenAI Batch API (async bulk processing), queries route directly to OpenAI. For real-time batch processing, send parallel requests through Fivo for maximum savings.

Question 14

What OpenAI rate limits apply with Fivo?

Accepted Answer

Fivo reduces the number of requests reaching OpenAI by 60-80%, so you are far less likely to hit OpenAI rate limits. If limits are hit, Fivo automatically manages retry and backoff. Enterprise plans include dedicated rate limit pools.

Question 15

Can Fivo route OpenAI queries to cheaper models?

Accepted Answer

Yes. Fivo can automatically route simple queries to GPT-3.5 Turbo, Claude Haiku, or Gemini Flash while keeping complex queries on GPT-4o. You set the quality threshold � Fivo minimizes cost.

Question 16

Does Fivo change OpenAI response format?

Accepted Answer

No. Responses are byte-for-byte compatible with the OpenAI API format. Your parsing code, type definitions, and error handling all work without changes.

Question 17

What happens if Fivo is down?

Accepted Answer

Fivo automatically falls back to direct OpenAI. Zero downtime for your application. You temporarily pay full OpenAI pricing until Fivo recovers. With 99.9% uptime (99.99% Enterprise), this is extremely rare.

Question 18

Does Fivo support OpenAI o1/o1-mini reasoning models?

Accepted Answer

Yes. Fivo supports o1 and o1-mini reasoning models. These models benefit from optimization for repeated reasoning patterns common in production workloads.

Question 19

Can I see which queries were optimized vs sent to OpenAI?

Accepted Answer

Yes. The Fivo dashboard shows real-time breakdown of optimized vs direct queries, with cost savings per query, latency comparisons, and model usage analytics.

Question 20

Does Fivo support the OpenAI Realtime API?

Accepted Answer

The Realtime API (WebSocket) is on the Fivo roadmap. Currently, Fivo optimizes REST-based Chat Completions and Embeddings endpoints which cover 95%+ of production use cases.

Fivo + OpenAI FAQ

Supported OpenAI Models

Model	Supported	Typical Savings
GPT-4o	Yes	100-5–20×
GPT-4 Turbo	Yes	100-5–20×
GPT-4	Yes	100-5–20×
GPT-3.5 Turbo	Yes	10-40x
o1 / o1-mini	Yes	50-150x
text-embedding-3-small	Yes	10x
text-embedding-3-large	Yes	10x
Fine-tuned models	Yes	50-5–20×