OpenRouter API Pricing Update: September 8, 2026
Por ahora este informe solo está disponible en inglés.
On September 8, 2026, OpenRouter tracks 428 models with 19 free models, including two removed from the free tier, and eight models with adjusted paid pricing.
Free Models
OpenRouter lists 19 free models as of September 8, 2026, including Google's Lyria 3 Pro Preview and Lyria 3 Clip Preview, both with 1,048,576 context tokens, and Thinking Machines' Inkling Small and Inkling, each with 1,048,576 context tokens.
NVIDIA offers four free models: Nemotron 3.5 Lightning and Nemotron 3 Ultra (1,000,000 context tokens), Nemotron 3 Super (262,144 context tokens), and Nemotron 3.5 Content Safety (128,000 context tokens).
Additional free models include inclusionAI's Ling 3.0 Flash Sante and Ling 3.0 Flash Fin (262,144 context tokens), Poolside's Laguna S 2.1 and Laguna XS 2.1 (262,144 context tokens), Google's Gemma 4 26B A4B and Gemma 4 31B (262,144 context tokens), Cohere's North Mini Code (256,000 context tokens), NVIDIA's Nemotron 3 Nano Omni (256,000 context tokens), the Free Models Router (200,000 context tokens), and LiquidAI's LFM2.5-2.6B (65,536 context tokens).
Changes to Free Tier
Two models were removed from the free tier compared to September 7, 2026: MiniMax M2.7 and MiniMax M3, both previously free and now no longer listed as free.
No models became free or stopped being free on this date beyond the removals; the became_free and stopped_being_free arrays are empty.
Paid Price Adjustments
Eight models experienced price changes between September 7 and September 8, 2026. Z.ai's GLM Flash Latest decreased input price from $0.075 to $0.0712 per million tokens and output from $0.25 to $0.2375 per million tokens.
DeepSeek V4 Flash Latest increased input from $0.045 to $0.05 per million tokens and output from $0.09 to $0.16 per million tokens. DeepSeek V4 Pro 0813 rose from $0.5795 to $0.66 input and $1.7384 to $1.98 output per million tokens.
Z.ai's GLM 4.6 reduced input from $0.55 to $0.43 and output from $2.2 to $1.75 per million tokens. Z.ai's GLM Latest lowered input from $1.17 to $1.12 and output from $3.96 to $3.52 per million tokens.
MoonshotAI Kimi Latest decreased input from $2.55 to $2.5 per million tokens but increased output from $12.75 to $14.0 per million tokens.
DeepSeek V4 Flash 0423 increased input from $0.0788 to $0.0886 and output from $0.1576 to $0.1772 per million tokens. DeepSeek V4 Pro 0423 rose sharply from $0.6167 to $0.9553 input and $1.2333 to $1.9105 output per million tokens.
Cheapest Paid Options
The lowest input cost among paid models is IBM's Granite 4.0 Micro at $0.017 per million tokens, with output at $0.112 per million tokens and 131,000 context tokens.
Mistral Nemo follows at $0.019 input and $0.03 output per million tokens with 131,072 context tokens. inclusionAI's Ling 3.0 Flash is third at $0.021 input and $0.063 output per million tokens with 262,144 context tokens.
Nex AGI's Nex-N2-Mini and OpenAI's GPT-5 Nano (batch) both charge $0.025 input per million tokens; Nex-N2-Mini outputs at $0.1 and GPT-5 Nano at $0.2, with context lengths of 262,144 and 400,000 tokens respectively.
Other low-cost options include Meta's Llama 3.2 1B Instruct at $0.027 input/$0.201 output (60,000 context), Upstage's Solar Pro 4 at $0.03 input/$0.12 output (524,288 context), and Qwen's Qwen3.7 Flash at $0.03 input/$0.13 output (1,000,000 context).
Puntos clave
- 19 models remain free on OpenRouter as of September 8, 2026, with two previously free MiniMax models removed from the free tier.
- Eight models had price updates, including decreases for Z.ai's GLM Flash Latest and GLM 4.6, and increases for DeepSeek V4 Pro variants.
- IBM's Granite 4.0 Micro is the cheapest paid model by input cost at $0.017 per million tokens.
- The lowest output cost among paid models is Mistral Nemo at $0.03 per million tokens.
- Context lengths for free models reach up to 1,048,576 tokens, while the largest windows overall are 2,000,000 tokens in router models.