Tryingopen API

Gemma 4 26B

by Google · 25B parameters

google/gemma-4-26b-a4b-it

The mixture-of-experts Gemma: only 3.8B of its 25B runs per word, so it is much quicker than its size suggests. Reads pictures and video. Through the API it takes text only.

Credits are charged at these prices whoever answers: the highest price any route lists for this model, including OpenRouter endpoints not shown below.

Providers

With no provider, requests go to Amazon Bedrock, with zero retention. Name one below to pin it; any provider preference goes through OpenRouter with zero retention.

Provider	Input / 1M	Output / 1M	Context	Zero retention	Slug to pin
Amazon Bedrock (default)	Not listed		256k	Yes	None: send no provider
CoreWeave	10	30	262k	Yes	coreweave
NextBit	9	30	262k	Yes	nextbit
DekaLLM	6	33	262k	Yes	dekallm
DeepInfra	7	34	262k	Yes	deepinfra
MaMakora	10	34	262k	Yes	makora
Novita	13	40	262k	Yes	novita
Parasail	13	40	262k	Yes	parasail
SiliconFlow	14	40	262k	Yes	siliconflow
Venice	13	40	256k	Yes	venice
GoGoogle	15	60	262k	Yes	google-vertex

Provider prices are their list prices, in credits, for comparison.

Code

curl https://www.tryingopen.com/v1/chat/completions \
  -H "Authorization: Bearer your-name" \
  -H "Content-Type: application/json" \
  -d '{"model":"google/gemma-4-26b-a4b-it","max_tokens":1000,"messages":[{"role":"user","content":"Hello!"}]}'

Try it

The API isn't set up on this deployment yet.