Frontier model performance at a fraction of the price
RouterPlus eval-based routing selects models for workloads using user-provided evals and production traffic.
routerplus.comCrawled Nov 16, 2026
RouterPlus eval-based routing selects models for workloads using user-provided evals and production traffic.
API pricing for 45 Particle, Ultramarine, Kimi, DeepSeek, GLM, Qwen, Gemma 4, Muse Spark, gpt-oss, MiniMax M3, Hy4 and Step 5 models across 18 providers, plus 5 Krea image models, with model IDs and context windows. One OpenAI-compatible key on RouterPlus.
RouterPlus shops model requirements across its provider network, offering committed capacity, quotes, or backup providers.
New models, providers and features on RouterPlus, newest first, with a link to each announcement.
A walkthrough of the RouterPlus console with a sample company, from the first request to budgets, guardrails, evals and coding agents.
RouterPlus retains request logs for up to 30 days on shared endpoints, not dedicated ones, and uses metadata-only billing logs with hashed API keys and encrypted credentials.
Anthropic launched Claude Fable 5.2 on Nov 9, 2026 at $8/$40 per million tokens, down from $10/$50, with 1M context and a $16/$80 fast mode.
Cloudflare deprecates 24 Workers AI models on November 20, 2026, including DeepSeek V4 Pro 0813, Kimi K2.6 and GLM 5.3 Flash. IDs and alternatives.
Gemini Omni 1.2 Flash is GA on the Gemini API: video output cut 30% to $12 per million tokens (about $0.07/s at 720p), with clips up to 15 seconds.
Parallel Decisions is in the RouterPlus catalog as parallel/decisions: web-grounded answers with citations, $0.001 per call for up to 100 questions.
Research, benchmarks, and product updates from the RouterPlus team.
StepFun's Step 5 Preview joined RouterPlus on Nov 10, 2026, served by RunInfra at $0.90 input and $2.43 output per million tokens with a 1M context.
The Claude API had elevated errors from 13:09 to 15:38 UTC on Nov 13, 2026, also hitting claude.ai and Claude Code. Timeline and what to check.
DeepSeek opened its app features to the API on Nov 13, 2026: web search, file input, URL reading, safety checks, plus OCR at $0.225 per 1,000 pages.
Google launched Gemini 4 Hydrogen (gemini-4-hydrogen) on Nov 12, 2026 at $1.50 input and $7.50 output per million tokens, with a free tier.
OpenAI now processes GPT-6 Luna API calls in India, Japan and Australia. Standard tier only, no hosted tools, 10% uplift: $0.11 in, $0.55 out.
Thinking Machines released Inkling 2 Small weights on Nov 12, 2026 under Apache 2.0. Together and Cloudflare list it at $0.30 input, $1.00 output per 1M.
Inference news for teams running models: launches, price changes, deprecations and API changes, with sources and RouterPlus catalog prices.
Meituan's LongCat-2.5, a 1.6T-parameter MoE with 48B active, is live on the LongCat API at a limited-time $0.30 input and $1.20 output per million tokens.
Lowest RouterPlus prices for 8 new open-weight models on Oct 31, 2026. Qwen3.8 27B from $0.039 input; GLM 5.3 Flash from $0.20 output per million tokens.
DeepSeek API pricing for 8 models on 16 providers: from $0.04 per million input tokens and $0.18 per million output tokens. Every provider's price, cached input rates and cost examples.
OpenAI released gpt-6.1-terra on Oct 26, 2026 at $1 input and $5 output per million tokens, with a 1.05M-token context and 128K max output.
The page compares RouterPlus and Cloudflare AI Gateway, focusing on provider pricing without a 5% credit fee, routing on user evals, reserved capacity, and sourced providers.
RouterPlus vs LiteLLM: a hosted, one-key alternative to running the LiteLLM proxy. Provider prices with no markup, 0% BYOK fee, reserved capacity. Checked Nov 2026.
The page compares RouterPlus and OpenRouter, highlighting routing on user evals, reserved capacity, negotiated token pricing, and a 0% BYOK fee.
The page compares RouterPlus and Portkey, emphasizing a single key for inference sales at provider prices, a 0% BYOK fee, reserved capacity, and sourced providers.
The page compares RouterPlus and Requesty, noting provider pricing without a 5% markup, reserved capacity across providers, and sourced providers.
The page compares RouterPlus and Vercel AI Gateway, noting routing on user evals, a 0% BYOK fee without credit purchases, reserved capacity, and sourced providers.
The page describes how to contact the RouterPlus team about sales, reserved capacity, support, or listing models as a provider.
RouterPlus routes AI inference and LLM workloads using user-provided evals and production traffic, managing capacity and access via a single key.