OpenRouter API Pricing Update: 20 Free Models, DeepSeek and Z.ai Price Shifts, New Kimi K3.1 Pro
OpenRouter tracks 469 models as of 2026-10-24, with 20 free models available and one new paid model, MoonshotAI: Kimi K3.1 Pro, entering at $2.00 per million input tokens and $10.00 per million output tokens.
Free model inventory
The free tier holds 20 models from eight providers. Google contributes four: Lyria 3 Pro Preview, Lyria 3 Clip Preview, Gemma 4 26B A4B, and Gemma 4 31B, all with 262,144 or 1,048,576 token contexts. NVIDIA provides six Nemotron variants spanning 128,000 to 1,000,000 token contexts. Poolside offers Laguna S 2.1 and Laguna XS 2.1 at 262,144 tokens. Thinking Machines lists Inkling Small and Inkling at 1,048,576 tokens. Cohere, XingChen, Dots Studio, Recursive, LiquidAI, and the OpenRouter Free Models Router each supply one model with contexts between 65,536 and 256,000 tokens.
New and removed models
MoonshotAI: Kimi K3.1 Pro is the only new model added since 2026-10-23. It carries a 1,048,576 token context, closed weights, and pricing of $2.00 per million input tokens and $10.00 per million output tokens. No models were removed.
Price changes
Twelve models updated pricing. Z.ai: GLM 5.3 Flash saw the largest input increase, from $0.0052 to $0.15 per million tokens, and output from $0.1985 to $0.50. Z.ai: GLM 5.3 dropped input from $0.70 to $0.079 and output from $2.20 to $1.40. Z.ai: GLM Flash Latest moved input from $0.0052 to $0.0304 and output from $0.1985 to $0.122. DeepSeek: DeepSeek Pro Latest cut input from $0.225 to $0.1188; output held at $8.00. DeepSeek: DeepSeek Flash Latest reduced input from $0.07 to $0.02 and raised output from $0.50 to $1.00. DeepSeek: DeepSeek V4 Flash 0731 lowered input from $0.44 to $0.0444 and output from $1.32 to $0.1333. DeepSeek: DeepSeek V4.1 Flash dropped input from $0.30 to $0.02; output from $1.20 to $1.00. DeepSeek: DeepSeek V4 Pro 0423 decreased input from $0.9553 to $0.6622 and output from $1.9105 to $1.3245. MoonshotAI: Kimi K3 kept input at $0.239 and cut output from $11.936 to $10.80. Google: Gemma 4 26B A4B trimmed input from $0.0765 to $0.0675 and output from $0.255 to $0.225. Mistral: Mistral Small 3.2 24B reduced input from $0.0938 to $0.075 and output from $0.25 to $0.20. Qwen: Qwen3.5-9B raised input from $0.08 to $0.17 and output from $0.13 to $0.25.
Cheapest paid models by input price
The eight lowest input prices start with IBM: Granite 4.0 Micro at $0.017 per million input tokens and $0.112 output, 131,000 token context, open weights. OpenAI: gpt-oss-20b follows at $0.018 input and $0.09 output, 131,072 tokens, open weights. DeepSeek: DeepSeek Flash Latest and DeepSeek: DeepSeek V4.1 Flash both sit at $0.02 input and $1.00 output with 1,048,576 token contexts; the former is closed weights, the latter open. inclusionAI: Ling 3.0 Flash offers $0.021 input and $0.063 output at 262,144 tokens, open weights. OpenAI: gpt-oss-20b (batch) at $0.024 input and $0.112 output, 131,072 tokens, open weights. Nex AGI: Nex-N2.5-Mini at $0.025 input and $0.10 output, 262,144 tokens, open weights. OpenAI: GPT-5 Nano (batch) at $0.025 input and $0.20 output, 400,000 tokens, closed weights.
Largest context windows
Five models share the maximum 2,000,000 token context: OpenRouter Auto Router (Beta), OpenRouter Pareto Code Router, SpaceXAI: Grok 4.20 Multi-Agent, SpaceXAI: Grok 4.20, and OpenRouter Auto Router.
Key takeaways
- 20 models remain free on OpenRouter, led by Google and NVIDIA with the highest context allowances.
- MoonshotAI: Kimi K3.1 Pro is the sole new model, priced at $2.00/$10.00 per million tokens with a 1M token context.
- DeepSeek models dominate the cheapest paid input tier, with four variants at or below $0.02 per million input tokens.
- Z.ai: GLM 5.3 Flash recorded the steepest input price increase, rising 28x from $0.0052 to $0.15 per million tokens.
- Five router or Grok models offer the platform's maximum 2M token context window.