AI API Providers
Compare pricing, models, and features across all 17 AI API providers in our database. Click any provider for a full pricing breakdown and calculator, or browse every model in the Pricing Index.
OpenAI
The industry leader. GPT-6 Astra, Sol and Luna lead, backed by the GPT-6.1, GPT-5.6 and GPT-5.4 families and budget nano options. Largest ecosystem and widest model selection.
Anthropic
Claude family — Fable 5.1, Opus 5.5 and Sonnet 5.5 — known for safety, tool use and long context. Haiku 5.5 for budget tasks.
Gemini family — the Gemini 4 Argon flagship plus Flash models that rank among the cheapest available, largest context windows (up to 1M tokens), and image models.
Mistral AI
European AI leader. Strong open-weight models — Large, Medium, Small, Codestral and Ministral — with data sovereignty focus.
DeepSeek
Ultra-budget Chinese AI. DeepSeek V4.1 Flash at $0.15/$0.60 off-peak is among the cheapest models available anywhere.
Cohere
Enterprise-focused with strong RAG and search capabilities. Command A for complex tasks, Command R+ and Command R for standard workloads.
Together.ai
Open-source models on optimized infrastructure. Llama 4 and Llama 3.3 at budget prices, with up to 1M context.
xAI
Grok family — Grok 4.7 flagship with real-time X data access, Grok 4.3 mid-tier, Grok Build 0.1 for budget.
Moonshot (Kimi)
Chinese AI startup. Kimi K3 flagship with 1M context, plus K2.6 and K2.7 Code budget models.
AI21 Labs
Hybrid SSM-Transformer architecture. Jamba 1.7 Large for enterprise with 256K context, Jamba Mini for budget.
Meta Model API and dynamic inference routing
Checked October 13, 2026: Muse Wonder (muse-wonder-1.0) is a public-preview Meta API model. Rates are $4 input, $0.20 cached input and $20 output per million tokens, with a 1,048,576-token context and per-team limits of 3,000 RPM / 4 million TPM. It accepts multimodal input and returns text; Meta lists no long-context premium. See the canonical Pricing Index and Meta pricing documentation.
Muse Wonder Full is restricted to eligible Muse Code subscriptions, not a public pay-as-you-go API. It has no calculator token price. For application integration rather than model rankings, read our Muse Wonder API integration guide.
DigitalOcean Inference Router is Public Preview: request inference-router with balanced (default), budget or quality. Chat Completions, Responses and Messages are supported. The serving model determines token charges; routing has no separate fee during preview, not a fixed token price. Legacy preset/custom slugs still resolve, while custom-router management remains available through the API rather than the removed dashboard controls. See DigitalOcean's current routing documentation.