Menu

13 tools · editorial

Best LLM APIs

Hosted language-model APIs for products that need to bring their own interface. StackPick ranks the API. OpenRouter rankings and Arena rank the models.

SDefault
90OpenAI APIProducts that want the broadest model, tool, and SDK support.Paid
AStrong
88Anthropic APIProduction apps that want Claude specifically, especially for long documents and code.Paid88Gemini APIApps that want Gemini, multimodal inputs, and a real free tier to prototype.Freemium87OpenRouterSwitching models without rewriting the client, or failing over when one lab is down.Freemium85DeepSeek APIHigh-volume workloads where token price matters and you can swap models.Paid85GroqChat and voice products where time-to-first-token is the constraint.Freemium81MistralTeams who want a European API vendor or to self-host Mistral weights.Freemium80Together AIServing and fine-tuning open weights without operating GPUs.Freemium80xAI APIProducts that specifically want Grok models behind their own UI.Paid
BSolid
79FireworksProduction traffic on open models when you want serverless, not a cluster.Freemium76Cerebras InferenceLow-latency open-model inference when you are already benchmarking Groq.Freemium75Amazon BedrockCompanies that already buy through AWS and need private networking.Paid73CohereRetrieval stacks that need embeddings and a reranker more than a celebrity chat model.Freemium
CSituational

Nothing in this tier.

Head-to-heads

OpenAI API vs Anthropic API

OpenAI API vs Gemini API

Groq vs OpenRouter

Questions

Which model API should you call?

OpenAI API is the top-ranked pick at 90. Products that want the broadest model, tool, and SDK support. The score is editorial, not a measured benchmark.

Which LLM APIs tools are free to start?

Gemini API (freemium), OpenRouter (freemium), Groq (freemium), Mistral (freemium), Together AI (freemium), Fireworks (freemium). Dollar prices say “check site” unless the software itself costs nothing to run. Confirm the current plan before you commit.

How does this ranking work?

Capability is 40% of the score, ease 25%, value 20%, and momentum 15%. S is 90 and above, A is 80 to 89, B is 70 to 79, and C is under 70. A sponsored label, if you see one, does not change the number.

How do OpenRouter rankings and Arena fit this list?

This page ranks API products, the tool you call. OpenRouter ranks models by tokens processed through the OpenRouter API. OpenRouter states that a higher token total shows how much a model is used, not which model is best for a task. That usage informs momentum (https://openrouter.ai/rankings). Arena, also called LMArena, ranks models by blind human votes. People compare anonymous answers, and only those anonymous votes count toward the public board. Arena fits the votes with a Bradley-Terry model. Those votes inform capability (https://arena.ai). Figures stay on those sites.