Does cheapest mean best?
No. Cheap models can be excellent for narrow workloads, but you should still check context, output limit, tool support, and quality signals before choosing one.
Cheapest LLM API models, input, output, context, and capability rankings
Low-cost LLM API routes ranked by input price, output price, 128K+ context, vision, tool calling, and JSON support.
Price-firstBy category
Input
Cheapest input
01
DeepSeek: DeepSeek V4 Flash (free)
OpenRouterDeepSeek
$0.00 / 1M
02
Google: Gemma 4 26B A4B (free)
OpenRouterGoogle
$0.00 / 1M
03
Google: Gemma 4 31B (free)
OpenRouterGoogle
$0.00 / 1M
04
Meta: Llama 3.2 3B Instruct (free)
OpenRouterMeta
$0.00 / 1M
05
Meta: Llama 3.3 70B Instruct (free)
OpenRouterMeta
$0.00 / 1M
06
OpenAI: gpt-oss-120b (free)
OpenRouterOpenAI
$0.00 / 1M
Output
Cheapest output
01
DeepSeek: DeepSeek V4 Flash (free)
OpenRouterDeepSeek
$0.00 / 1M
02
Google: Gemma 4 26B A4B (free)
OpenRouterGoogle
$0.00 / 1M
03
Google: Gemma 4 31B (free)
OpenRouterGoogle
$0.00 / 1M
04
Meta: Llama 3.2 3B Instruct (free)
OpenRouterMeta
$0.00 / 1M
05
Meta: Llama 3.3 70B Instruct (free)
OpenRouterMeta
$0.00 / 1M
06
OpenAI: gpt-oss-120b (free)
OpenRouterOpenAI
$0.00 / 1M
Context
Cheapest 128K+ context
01
DeepSeek: DeepSeek V4 Flash (free)
OpenRouterDeepSeek
$0.00 / 1M
02
Google: Gemma 4 26B A4B (free)
OpenRouterGoogle
$0.00 / 1M
03
Google: Gemma 4 31B (free)
OpenRouterGoogle
$0.00 / 1M
04
Meta: Llama 3.2 3B Instruct (free)
OpenRouterMeta
$0.00 / 1M
05
OpenAI: gpt-oss-120b (free)
OpenRouterOpenAI
$0.00 / 1M
06
OpenAI: gpt-oss-20b (free)
OpenRouterOpenAI
$0.00 / 1M
Vision
Cheapest vision
01
Google: Gemma 4 26B A4B (free)
OpenRouterGoogle
$0.00 / 1M
02
Google: Gemma 4 31B (free)
OpenRouterGoogle
$0.00 / 1M
03
Free Models Router
OpenRouterOther
$0.00 / 1M
04
NVIDIA: Nemotron 3 Nano Omni (free)
OpenRouterOther
$0.00 / 1M
05
NVIDIA: Nemotron Nano 12B 2 VL (free)
OpenRouterOther
$0.00 / 1M
06
Google: Gemma 3 12B
OpenRouterGoogle
$0.0400 / 1M
Tools
Cheapest tools
01
DeepSeek: DeepSeek V4 Flash (free)
OpenRouterDeepSeek
$0.00 / 1M
02
Google: Gemma 4 26B A4B (free)
OpenRouterGoogle
$0.00 / 1M
03
Google: Gemma 4 31B (free)
OpenRouterGoogle
$0.00 / 1M
04
Meta: Llama 3.3 70B Instruct (free)
OpenRouterMeta
$0.00 / 1M
05
OpenAI: gpt-oss-120b (free)
OpenRouterOpenAI
$0.00 / 1M
06
OpenAI: gpt-oss-20b (free)
OpenRouterOpenAI
$0.00 / 1M
JSON
Cheapest JSON
01
Google: Gemma 4 26B A4B (free)
OpenRouterGoogle
$0.00 / 1M
02
Google: Gemma 4 31B (free)
OpenRouterGoogle
$0.00 / 1M
03
Free Models Router
OpenRouterOther
$0.00 / 1M
04
MiniMax: MiniMax M2.5 (free)
OpenRouterOther
$0.00 / 1M
05
NVIDIA: Nemotron 3 Super (free)
OpenRouterOther
$0.00 / 1M
06
NVIDIA: Nemotron Nano 9B V2 (free)
OpenRouterOther
$0.00 / 1M
FAQ
About choosing low-cost models
Why do free models appear in the rankings?
Free route prices are public pricing records. Unit-priced routes are excluded from token-price rankings so they do not appear as free token models.
Why are input and output rankings separate?
Some workloads are input-heavy while others generate long outputs. Separate rankings make it easier to match model cost to the shape of your usage.
