DeepSeek V4.1 Flash input price drops to $0.0507 per million
DeepSeek V4.1 Flash’s input cost is now $0.0507 per million tokens, down from $0.30, marking the steepest input‑price reduction in today’s OpenRouter snapshot.
Free Model Landscape
OpenRouter tracks 471 models, of which 17 are free and 191 are open‑weights. The free set includes high‑context offerings such as inclusionAI: Ling 3.2 Flash (free) with 1,048,576 tokens, Particle AI: Particle-1 (free) with the same window, and Google’s Lyria 3 Pro Preview also at 1,048,576 tokens. Smaller windows appear in models like Dots Studio: Dots3-Note Preview (free) at 512,000 tokens and LiquidAI: LFM3-1.7B (free) at 262,144 tokens. The OpenRouter "Free Models Router" provides a baseline of 200,000 tokens for routing purposes.
Price Adjustments
Particle AI: Particle-1 raised its input price from $0.05 to $0.06 per million tokens while output remains $0.5. Z.ai: GLM 5.2 lowered input from $0.322 to $0.298 and output from $0.989 to $0.916. OpenAI: GPT-5.6 Sol cut input from $4.4 to $4.0 and output from $22.0 to $20.0. DeepSeek: R1 0528 increased input from $0.5 to $0.571 and output from $2.15 to $2.286. Qwen: Qwen3.5-9B reduced input dramatically from $0.156 to $0.039 and output from $0.229 to $0.058. DeepSeek: DeepSeek V4.1 Flash dropped input from $0.3 to $0.0507 and output from $1.2 to $0.2028.
Cheapest Paid Options
Among paid models, Z.ai: GLM 5.2 leads with an input price of $0.007 per million tokens and an output price of $6.8, supporting 1,048,576 tokens and flagged as open‑weights. inclusionAI: Ling 3.0 Tiny follows at $0.015 input and $0.045 output for 262,144 tokens, also open‑weights. Meta: Llama 3.1 8B Instruct offers $0.02 input and $0.04 output on a 131,072‑token window, open‑weights. inclusionAI: Ling 3.0 Flash is priced at $0.021 input and $0.063 output for 262,144 tokens, open‑weights. OpenAI: gpt-oss-20b (batch) costs $0.024 input and $0.112 output on 131,072 tokens, open‑weights. Nex AGI: Nex-N2.5-Mini charges $0.025 input and $0.1 output for 262,144 tokens, open‑weights. OpenAI: GPT-5 Nano (batch) is $0.025 input and $0.2 output on 400,000 tokens, not open‑weights. Meta: Llama 3.2 1B Instruct rounds out the list at $0.027 input and $0.201 output for 60,000 tokens, open‑weights.
Largest Context Windows
Five entries provide the maximum 2,000,000 token context window: SpaceXAI: Grok 4.20 Multi-Agent, SpaceXAI: Grok 4.20, and three OpenRouter services—Auto Router, Auto Code, and Auto Router (Beta). These models enable the longest prompt lengths available on the platform today.
Key takeaways
- DeepSeek V4.1 Flash now has the lowest input price among the models that changed today.
- Qwen3.5-9B also saw large input and output price cuts, reaching $0.039 and $0.058 per million tokens respectively.
- The free model count remains at 17 and open‑weight models stay at 191, unchanged from yesterday.
- Five models share the top context window of 2,000,000 tokens across providers.