NOUMENOpen app
NOUMEN

Google · Open weights

Gemma 4 26B A4B Thinking

Added Apr 2, 2026Open weights

Compare models

Choose up to 3 models to compare.1 of 3 chosen

Loading models…

Use in the app

Google's Gemma 4 26B A4B instruction-tuned model with structured reasoning for more deliberate coding, multimodal analysis, and long-context problem solving. Requests containing video cost $0.10 per million input tokens and $0.40 per million output tokens.

  • Reasoning
  • Tools
  • Vision
  • Structured output
  • Video input
Input
Text, Images, Video
Output
Text
Discovery tier
Open weights

Pricing

USD per 1M tokens
Input
$0.09
Output
$0.30
Cache read
$0.05

Reference prices as published by the model provider. API example

Specifications

Context window
262.1K
Max output
131.1K
Average output (7 days)
1.1K tokens
Longer than 73% of 468 models

Side by side

Gemma 4 26B A4B Thinking compared

Specifications, prices, speed and benchmark scores next to each other. Benchmarks below show every compared model too.

Compare models

Choose up to 3 models to compare.1 of 3 chosen

Loading models…

Gemma 4 26B A4B ThinkingGoogleThis modelGemma 4 12B InstructGoogleRemove
ProviderGoogleGoogle
Discovery tierOpen weightsOpen weights
AddedApr 2, 2026Aug 1, 2026
Limits
Context window262.14KBest131.07K
Max output131.07KBest32.77K
Average output1.1K tokens162 tokens
Pricing, USD per 1M tokens
Input$0.09$0.05Best
Output$0.30$0.25Best
Cache read$0.05$0.025Best
Performance
Throughput31.8 tok/sBest23.5 tok/s
Time to first token4.4 s2.5 sBest
Request success (24h)100.0%Best98.6%
Benchmark indexes
Intelligence Index16.7Best9.4
Coding Index39.3—
Capabilities
ReasoningYesNo
ToolsYesYes
VisionYesYes
Structured outputYesYes
Video inputYesNo
Reasoning effort——

Public evaluations

Benchmarks

Independent evaluation results, with how each score ranks among the models in the Noumen catalog.

  • Gemma 4 26B A4B Thinking
  • Gemma 4 12B Instruct

Intelligence Index

  • Gemma 4 26B A4B Thinking16.7
  • Gemma 4 12B Instruct9.4

Gemma 4 26B A4B Thinking: Better than 44% of 326 scored models

3.8Catalog models by score55.1

Indexes

  • Coding Index

    Composite of the coding evaluations

    • Gemma 4 26B A4B Thinking: 39.3
    • Gemma 4 12B Instruct: —

Agentic work

  • AutomationBench-AA

    Workflow automation, penalized for guardrail violations

    • Gemma 4 26B A4B Thinking: 1.9%
    • Gemma 4 12B Instruct: —

  • AutomationBench-AA tasks completed

    Workflows finished end to end without a guardrail violation

    • Gemma 4 26B A4B Thinking: 0.2%
    • Gemma 4 12B Instruct: —

  • AA-Briefcasev1.1

    Agentic knowledge work, rated head to head

    • Gemma 4 26B A4B Thinking: 315 Elo
    • Gemma 4 12B Instruct: —

  • GDPval-AAv2.1

    Economically valuable occupational tasks, rated head to head

    • Gemma 4 26B A4B Thinking: 565 Elo
    • Gemma 4 12B Instruct: —

Document reasoning

  • AA-LCRv1.1

    Reasoning across long documents

    • Gemma 4 26B A4B Thinking: 65.7%Top
    • Gemma 4 12B Instruct: 35.0%

Reasoning and knowledge

  • Humanity’s Last Exam

    Expert-level questions across academic fields

    • Gemma 4 26B A4B Thinking: 19.3%Top
    • Gemma 4 12B Instruct: 6.3%

  • GPQA Diamond

    Graduate-level science questions

    • Gemma 4 26B A4B Thinking: 79.2%Top
    • Gemma 4 12B Instruct: 66.1%

  • IFBench

    Following precise output instructions

    • Gemma 4 26B A4B Thinking: 72.4%Top
    • Gemma 4 12B Instruct: 45.2%

Coding

  • SciCode

    Scientific computing in Python

    • Gemma 4 26B A4B Thinking: 40.0%
    • Gemma 4 12B Instruct: —

Legacy evaluations

  • Terminal-Bench Hard

    Earlier terminal task suite

    • Gemma 4 26B A4B Thinking: 13.6%Top
    • Gemma 4 12B Instruct: 11.4%

  • τ²-Bench Telecom

    Tool use in a customer-support simulation

    • Gemma 4 26B A4B Thinking: 43.6%Top
    • Gemma 4 12B Instruct: 31.9%

  • AA-LCR (unversioned)

    Long-context reasoning, earlier grading

    • Gemma 4 26B A4B Thinking: 65.7%Top
    • Gemma 4 12B Instruct: 35.0%

Scores from Artificial Analysis, updated Oct 26, 2026. Artificial Analysis

Live routing data

Performance and reliability

Recent averages across requests routed for this model. Your own speed depends on prompt length, reasoning effort and load.

Throughput
31.8 tok/s
Faster than 22% of 254 models
Time to first token
4.4 s
Quicker than 42% of 254 models
Prompt cache hits
0.5%
About $0.0898 per 1M input tokens with caching
Request success (24h)
100.0%
Reliable

Cache hits are the share of input tokens recently served from a prompt cache across all requests. They are not a forecast of your own hit rate.

Hourly request success · last 72 hoursLowest hour 96.1%
  • Healthy (98% or more)
  • Degraded (90–98%)
  • Down (under 90%)
  • No requests

Developers

Use it through the API

Noumen’s API is OpenAI-compatible: any Chat Completions client works with a base URL and a key.

Read the quickstart
curl https://www.noumen.xyz/api/v1/chat/completions \
  -H "Authorization: Bearer $NOUMEN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "google/gemma-4-26b-a4b-it:thinking",
    "messages": [{ "role": "user", "content": "Explain this stack trace." }]
  }'
import os
from openai import OpenAI

client = OpenAI(base_url="https://www.noumen.xyz/api/v1", api_key=os.environ["NOUMEN_API_KEY"])
reply = client.chat.completions.create(
    model="google/gemma-4-26b-a4b-it:thinking",
    messages=[{"role": "user", "content": "Explain this stack trace."}],
)
print(reply.choices[0].message.content)

Google

Other Google models, starting with the same family.