OpenAI API alternatives (2026): 8 options compared — vendors, gateways, and what stops working

There are two kinds of OpenAI API alternative: another model vendor's own API, or an OpenAI-compatible endpoint that keeps your SDK and changes only base_url and the key. The list below gives 8 options — four model vendors, two gateways and two self-hosted servers — with the protocol, the models, how each bills and its documentation, checked on October 22, 2026. Kunavo is one of the gateways, and it is our product.

“Alternative to the OpenAI API” is two different swaps, and most roundups on this search answer only one of them. Swapping the model — to Claude, to Gemini, to open weights — changes providers, accounts, SDKs and billing. Swapping the endpoint changes a base_url and an API key and leaves your code alone, because the OpenAI wire format is the thing being copied. Pick which swap you are making first; a list of models cannot tell you what the migration costs.

The question the lists skip is the one that decides it: which OpenAI surfaces actually survive the move. This page answers that for Kunavo specifically, including the three rows where the answer is no. Bias declared: Kunavo is our product — an independent OpenAI-compatible gateway, not OpenAI and not Anthropic. Rates and availability last verified October 23, 2026.

8 OpenAI API alternatives, side by side

Each entry below states only what the provider's own documentation said on October 22, 2026. The order is not a ranking, and no entry is a supplier of Kunavo's.

1. Anthropic Claude API

Anthropic's own API for the Claude models. The native interface is the Messages API. Anthropic also offers an OpenAI SDK compatibility layer at https://api.anthropic.com/v1/, which it describes as primarily for testing and comparing models rather than a long-term production solution; prompt caching is not supported on it. Billed per token at Anthropic's list prices. Documentation.

2. Google Gemini API

Google's API for the Gemini models, with an OpenAI compatibility endpoint at https://generativelanguage.googleapis.com/v1beta/openai/ that Google describes as beta. Google lists a free tier with free input and output tokens on some models, and a paid tier with higher rate limits. Documentation.

3. DeepSeek API

DeepSeek's API for its own models, in both the OpenAI and the Anthropic formats (https://api.deepseek.com and https://api.deepseek.com/anthropic). Usage is billed per token and deducted from a topped-up or granted balance. Documentation.

4. xAI API

xAI's API for the Grok models at https://api.x.ai/v1; its quick start uses the OpenAI SDK with xAI's base URL and key. Prices are per million tokens. Documentation.

5. OpenRouter

A gateway to hundreds of models from many vendors behind one OpenAI-compatible endpoint, https://openrouter.ai/api/v1. Its FAQ says credits are bought in advance with a fee on the purchase and no markup on inference, and that some models are free with low daily rate limits. Documentation.

6. Kunavo

Our product: an independent gateway with one key for Claude, GPT, image and video models, on the OpenAI chat completions and Responses formats and on Anthropic's Messages API, at https://api.kunavo.com/v1. Text models are billed per token, image models per image and video models per video or per second, at the rates on each model's page; there is no free tier. Which OpenAI surfaces survive the switch is in the table further down. Documentation.

7. Ollama

Runs open-weight models on your own machine and serves them on an OpenAI-compatible endpoint at http://localhost:11434/v1/: chat completions, completions, embeddings, models and a non-stateful Responses API. The SDK needs a key, which the local server ignores. The software is free; you pay for the hardware. Documentation.

8. vLLM

An open-source inference server: vllm serve with a model name exposes OpenAI-compatible Chat Completions, Responses and Embeddings routes, and audio transcription for speech models. Free software on your own GPUs. Documentation.

The three routes, and what each one costs you in code

A provider's own compatibility endpoint is a fourth, narrower option: Google publishes an OpenAI-compatibility layer for the Gemini API, which is the same base_url move confined to one vendor's models.

Swapping the endpoint, in full

# The whole migration, when you swap the endpoint rather than the model.
from openai import OpenAI

client = OpenAI(
    api_key="sk-kn-...",                      # was sk-proj-...
    base_url="https://api.kunavo.com/v1",     # the only other line that moves
)

# Same SDK, same call shape, same streaming, same tool-call format.
resp = client.chat.completions.create(
    model="claude-sonnet-5",                  # now reachable on the same key
    messages=[{"role": "user", "content": "Summarise this changelog."}],
    stream=True,
)

The full walk-through, from creating a key to the first request, is the quickstart.

Which OpenAI surfaces survive the swap

Read the middle column before migrating. “Route only” means the wire format is implemented and no model is enabled behind it, so the call fails — that is a worse surprise than an unimplemented route, and it is why the row exists rather than being quietly omitted.

The 3 “route only” rows are the ones to plan around. A RAG pipeline is the common case: embed against an embedding provider directly, generate against the gateway. The two are separate HTTP calls either way, so the cost is a second key rather than a second architecture. GET /v1/models is always the authority on what is callable today.

What the swap costs per token

Comparing like for like on Kunavo as of September 2026: Claude Sonnet 4.6 at $2.10 in / $10.50 out per 1M tokens against Anthropic's $3.00 / $15.00, and GPT-5.6 Sol at $2.00 / $12.00 against the promotional $4.00 / $20.00 OpenAI is charging now (list $5.00 / $30.00; the promotion runs at least through November 21, 2026). Check OpenAI's side on their pricing page rather than taking ours for it. The lever that usually moves a bill more than the vendor choice is prompt caching: an agent re-sends a large system prompt every turn, so uncached input dominates.

What Kunavo is not

Not OpenAI, and not a way to get OpenAI's hosted agent infrastructure somewhere cheaper — Assistants, Threads, Vector Stores, Realtime, Batch, fine-tuning and moderations are not implemented, and an app built on those is not a base_url away from moving. Not open-weight hosting either: if the answer you want is Llama or Qwen on your own hardware, a self-hosted OpenAI-compatible server is the route and this is not it. And not an embeddings provider, per the table above.

For the neighbouring comparisons: OpenRouter alternatives, LiteLLM alternatives for a proxy you run yourself over your own provider keys, the four kinds of LLM gateway, and the chat endpoint reference for the exact request and response shapes.

Frequently asked questions