LLMCost

Free LLM API models (18)

Every model below costs $0 for both input and output tokens via OpenRouter. Perfect for prototyping, hobby projects, and high-volume low-stakes workloads. Free tiers usually have rate limits and may use your prompts for training โ€” check each provider's terms. Verified 2026-08-23.

ModelContextMax outputPrice
NVIDIA: Nemotron 3.5 Lightning (free)
nvidia/nemotron-3.5-lightning:free
1M66KFREE
NVIDIA: Nemotron 3 Ultra (free)
nvidia/nemotron-3-ultra-550b-a55b:free
1M66KFREE
Dots Studio: Dots3-Note Preview (free)
dots-studio/dots-3-note-preview:free
512K512KFREE
Thinking Machines: Inkling Small (free)
thinkingmachines/inkling-small:free
262K262KFREE
Poolside: Laguna S 2.1 (free)
poolside/laguna-s-2.1:free
262K33KFREE
Thinking Machines: Inkling (free)
thinkingmachines/inkling:free
262K262KFREE
Poolside: Laguna XS 2.1 (free)
poolside/laguna-xs-2.1:free
262K33KFREE
Google: Gemma 4 26B A4B (free)
google/gemma-4-26b-a4b-it:free
262K33KFREE
Google: Gemma 4 31B (free)
google/gemma-4-31b-it:free
262K33KFREE
NVIDIA: Nemotron 3 Super (free)
nvidia/nemotron-3-super-120b-a12b:free
262K262KFREE
Cohere: North Mini Code (free)
cohere/north-mini-code:free
256K64KFREE
Z.ai: GLM 5.2 (free)
z-ai/glm-5.2:free
256K256KFREE
NVIDIA: Nemotron 3 Nano Omni (free)
nvidia/nemotron-3-nano-omni-30b-a3b-reasoning:free
256K66KFREE
NVIDIA: Nemotron 3 Nano 30B A3B (free)
nvidia/nemotron-3-nano-30b-a3b:free
256Kโ€”FREE
NVIDIA: Nemotron 3.5 Content Safety (free)
nvidia/nemotron-3.5-content-safety:free
128K8KFREE
NVIDIA: Nemotron Nano 12B 2 VL (free)
nvidia/nemotron-nano-12b-v2-vl:free
128K128KFREE
NVIDIA: Nemotron Nano 9B V2 (free)
nvidia/nemotron-nano-9b-v2:free
128Kโ€”FREE
LiquidAI: LFM2.5-2.6B (free)
liquid/lfm-2.5-2.6b:free
66K8KFREE

When should you use free models?

  • Prototyping โ€” validate your app before spending a cent
  • High-volume classification/extraction โ€” where quality bar is low
  • Learning & experiments โ€” zero-risk API practice

When you outgrow free tiers, check the cost calculator to find the cheapest paid model for your workload โ€” often just a few dollars a month.