Get alerts
All products/LLM APIs/OpenRouter

OpenRouter pricing, price hikes and cheaper alternatives

OpenRouter (openrouter.ai) is a LLM API. Usage-based per token at each route's listed price, with no subscription, plus a 5.5% fee on credit purchases (8% on Business).. Free plan: Free models: 50 requests/day (20/min), or 1,000/day after buying $10+ of credits; production-use terms not stated..

What OpenRouter costs

  • Top model: $211/mo (Claude Fable 5 (anthropic/claude-fable-5), ranked #1 at 1506 Elo on the LMArena text board in an Aug 12, 2026 model-ranking snapshot found via secondary sites. LMArena (lmarena.ai and arena.ai) returned HTTP 403, so the current rank is unverified; Claude Fable 5.1 has the same price and no text-board rank was found for it. Price $10 in / $50 out per 1M on its cheapest route (Anthropic, Bedrock, Azure, Google): 10 x 10 + 2 x 50 = $200; x 1.055 = $211.00.)
  • Cheapest model: $0.22/mo (mistralai/mistral-nemo via the Io Net route: $0.01656 in / $0.0216 out per 1M, 128,000-token context (counted as 128K). 10 x 0.01656 + 2 x 0.0216 = $0.2088; x 1.055 = $0.2203. Every paid route of the 270 live tool-calling text models with at least 128K context was checked (1,191 routes, 1,125 qualifying); :free, :batch, :flex and models with announced retirement dates are excluded. If 128K is read as 131,072 tokens, Io Net drops out and the budget becomes deepseek/deepseek-v4-flash via Baidu at $0.2534 x 1.055 = $0.2667.)
  • gpt-oss-120b: $0.63/mo (openai/gpt-oss-120b via Near AI (fp4, 131,072-token context, tool calling): $0.03 in / $0.15 out per 1M. 10 x 0.03 + 2 x 0.15 = $0.60; x 1.055 = $0.633. Venice ties on price (128,000 context). The model-level list price ($0.15 / $0.60, $2.70 workload) is not the cheapest route, and the :batch variant is excluded.)

Plan compared: Standard plan: cheapest route list price per 1M tokens for each point, on 10M input and 2M output tokens a month, plus t. Monthly USD cost of one reference workload: 10,000 requests a month, totaling 10M input and 2M output tokens, at standard pay-as-you-go rates with no caching or batch discounts. Top model is the product's highest-ranked model on the LMArena Text leaderboard. Cheapest model is its lowest-cost model with tool calling and a 128K context. gpt-oss-120b compares hosts on the same open model.

What OpenRouter does

Core: Multi-lab model catalog, OpenAI-compatible endpoint, Cross-provider failover.

Important: Frontier-grade models, Open-weight models (partly), Image and PDF input, 1M-token context, Hosted agent tools, Prompt caching discounts, Ongoing free tier, No-training, zero retention.

Also has: Ultra-fast inference (partly), Voice and realtime audio (partly), Image and video generation (partly), Embeddings and reranking, Batch and flex pricing (partly).

OpenRouter price hikes

No price hikes on record for OpenRouter.

Cheaper alternatives to OpenRouter

At 12M tokens a month on its top model, OpenRouter costs $211/mo. 2 LLM APIs cost less:

  • Gemini API: $40/mo (Pay-as-you-go Standard paid tier, text only, list prices as of 2026-10-17), covers 70% of OpenRouter; missing or partial: Multi-lab model catalog, OpenAI-compatible endpoint (partly), Cross-provider failover, No-training, zero retention (partly), Embeddings and reranking (partly)
  • OpenAI API: $200/mo (Pay-as-you-go per token, GPT-6 Astra at standard rates (flagship proxy)), covers 65% of OpenRouter; missing or partial: Multi-lab model catalog, Cross-provider failover, Ongoing free tier, No-training, zero retention (partly), Ultra-fast inference

OpenRouter alternatives: 3 LLM APIs compared, at every size.

Other kinds of products that cover part of OpenRouter

  • Hugging Face (GPU cloud & model hosting): covers 58% of OpenRouter; missing or partial: Frontier-grade models (partly), Image and PDF input (partly), 1M-token context (partly), Hosted agent tools, Prompt caching discounts
  • Perplexity (AI assistants): covers 48% of OpenRouter; missing or partial: OpenAI-compatible endpoint (partly), Cross-provider failover (partly), Image and PDF input, 1M-token context (partly), Prompt caching discounts

Every change we logged

  • Oct 12, 2026 · Price cut: Temporary 50% discount on all ElevenLabs models through OpenRouter, valid until 2026-10-19.
  • Oct 5, 2026 · Repriced: Enterprise BYOK allowance changed from $200,000/month (no fee within it) to custom terms; spend alerts added for Business and Enterprise. Direction unclear.
  • Sep 14, 2026 · Repriced: Business tier added between Standard and Enterprise with an 8% platform fee on credit purchases. A new higher tier, not a price hike; Standard stays at 5.5%.
  • Jul 6, 2026 · Repriced: BYOK free allowance moved from request counts to list-price spend thresholds: $25,000/month on Standard and Business, $200,000/month on Enterprise, then a 5% fee above each. Direction for customers is unclear.

Sources

Pricing page: openrouter.ai/pricing. Prices last checked Oct 16, 2026. Budget and oss come from OpenRouter's public models and endpoints APIs and the fee from openrouter.ai/pricing (all checked 2026-10-17); PricingSaaS tracking starts 2026-05-16, so no earlier changes are recorded and none of the logged changes is a clear price increase. The top point uses Claude Fable 5, the #1 text-board entry in Aug 2026 secondary snapshots, because lmarena.ai and arena.ai both returned HTTP 403; Fable 5.1 has the same $10/$50 price.