Gemma 4 26B A4B

Gemma 4 26B A4B by Google DeepMind: League Score, pricing, context window and sourced benchmarks — the full tracked record.

40 Open weights 262.1K context

League positions: #34 overall · #18 Reasoning Division · #13 Open-Weight Division. Score coverage: 15% of components measured. Count tokens & cost for Gemma 4 26B A4B · Estimate a workload cost

Agent-ready: this model is tracked with agentic strengths. See the AI Agents section for the products and frameworks that put models like this to work.

Specifications & pricing

Maker	Google DeepMind — lab profile
Family	Gemma 4
Released	Apr 2, 2026
Status	Current
Weights	Open weights
License	Apache-2.0
Knowledge cutoff	Not published
Context window	262,144 tokens
Max output	—
Modalities in	text, image
Modalities out	text
API access	Hugging Face, Google AI Studio

Benchmark results

Benchmark	Score	Source	Date	Link
GPQA Diamond	82.3 / 100	BenchLM GPQA leaderboard	Oct 14, 2026	Source

How its League Score is built

Component	Raw	Normalised	Weight	Date	Link
GPQA Diamond	82.3	40.4	15%	Oct 14, 2026	Source

Normalised 0–100 against the tracked field; missing components redistribute their weight. Full rules on the methodology page.

Closest rivals

Gemma 4 26B A4B vs Qwen3-Coder-480B-A35B-Instruct — full comparison or compare any pair
Gemma 4 26B A4B vs Inkling — full comparison or compare any pair
Gemma 4 26B A4B vs Llama 4 Maverick — full comparison or compare any pair

FAQ

How much does Gemma 4 26B A4B cost?

Gemma 4 26B A4B has no first-party API price in our database — it is distributed as open weights or via third-party hosts.

What is the context window of Gemma 4 26B A4B?

Gemma 4 26B A4B accepts up to 262,144 tokens of context.

Is Gemma 4 26B A4B open source?

Yes — its weights are publicly downloadable (license: Apache-2.0).

Data verified Oct 14, 2026. Spotted an error? Every figure above links to its source.