LLM API Pricing Update: 19 Free Models, 10 Price Changes, and Cheapest Paid Options on OpenRouter
As of 2026-10-30, OpenRouter tracks 456 models with 19 free options available, 10 models saw price changes from the previous day, and 1 model was removed from the catalog.
Free Models Snapshot
Nineteen models are currently free on OpenRouter, spanning providers including Google, NVIDIA, Thinking Machines, Dots Studio, Recursive, XingChen, Poolside, Cohere, LiquidAI, and OpenRouter itself.
Google leads with four free models: Lyria 3 Pro Preview, Lyria 3 Clip Preview, Gemma 4 26B A4B, and Gemma 4 31B, all with 1,048,576 or 262,144 token context windows.
NVIDIA offers five free models: Nemotron 3.5 Super VL, Nemotron 3.5 Lightning, Nemotron 3 Ultra, Nemotron 3 Nano Omni, and Nemotron 3.5 Content Safety, with context windows ranging from 128,000 to 1,000,000 tokens.
The Free Models Router from OpenRouter provides a 200,000 token context window, while LiquidAI's LFM2.5-2.6B offers the smallest free context at 65,536 tokens.
Models Removed and Added
One model was removed from the OpenRouter catalog: Upstage: Solar Pro 3 from provider Upstage.
No new models were added to the catalog on 2026-10-30.
No models transitioned to free status or stopped being free on this date.
Price Changes Across 10 Models
DeepSeek: DeepSeek Flash Latest saw input prices rise from $0.0138 to $0.0339 per million tokens, while output prices fell from $1.80 to $1.14 per million tokens.
DeepSeek: DeepSeek V4 Pro 0423 increased both input and output prices, from $0.9553 to $1.60 and from $1.9105 to $3.20 per million tokens respectively.
MoonshotAI: Kimi Latest reduced input costs from $0.1824 to $0.1624 per million tokens but raised output costs from $9.3388 to $9.7471 per million tokens.
Mistral: Mistral Large 4 cut prices significantly, with input dropping from $1.36 to $0.70 and output from $4.18 to $2.10 per million tokens.
Z.ai adjusted four models: GLM Flash Latest input rose from $0.0379 to $0.0503 and output fell from $0.0823 to $0.0701; GLM 5.3 Flash input rose from $0.0379 to $0.15 and output from $0.0823 to $0.50; GLM 4.6 input fell from $0.55 to $0.50 and output from $2.20 to $2.00; GLM 5.2 input held at $0.006 but output rose from $5.40 to $7.00 per million tokens.
DeepSeek: DeepSeek V4 Flash Latest increased input from $0.0475 to $0.056 and output from $0.1568 to $0.30 per million tokens.
MoonshotAI: Kimi K3 raised input from $0.192 to $0.239 and cut output from $20.00 to $10.80 per million tokens.
Cheapest Paid Models by Input Price
Z.ai: GLM 5.2 remains the cheapest paid model by input price at $0.006 per million tokens, though its output price of $7.00 per million tokens is high.
OpenAI: gpt-oss-20b follows at $0.018 input and $0.09 output per million tokens, with a 131,072 token context window and open weights.
inclusionAI: Ling 3.0 Flash costs $0.021 input and $0.063 output per million tokens with a 262,144 token context window and open weights.
OpenAI: gpt-oss-20b (batch) is priced at $0.024 input and $0.112 output per million tokens, while Nex AGI: Nex-N2.5-Mini costs $0.025 input and $0.10 output per million tokens, both with open weights.
Meta: Llama 3.2 1B Instruct and Mistral: Mistral Nemo round out the list at $0.027 and $0.029 input per million tokens respectively, both with open weights.
Largest Context Windows
Five models offer the largest context windows at 2,000,000 tokens: Auto Router (Beta), Pareto Code Router, SpaceXAI: Grok 4.20 Multi-Agent, SpaceXAI: Grok 4.20, and Auto Router, all from OpenRouter or x-ai.
These models provide the maximum context length available on the platform as of 2026-10-30.
Key takeaways
- Nineteen models are free on OpenRouter as of 2026-10-30, with Google and NVIDIA contributing the most free options.
- Ten models experienced price changes on 2026-10-30, with DeepSeek, MoonshotAI, Mistral, and Z.ai all adjusting their rates.
- Z.ai: GLM 5.2 is the cheapest paid model by input price at $0.006 per million tokens, despite a high output cost of $7.00.
- Upstage: Solar Pro 3 was the only model removed from the OpenRouter catalog on this date.
- Five models offer 2,000,000 token context windows, the largest available on the platform.