Gemma 4 26B A4B IT

This Simplismart deployment-focused article targets the Gemma 4 26B MoE variant by name, framing it as delivering near-flagship 31B-level performance at roughly 4B-parameter active inference cost. The piece discusses the MoE routing mechanism, contrasts the 26B-A4B against the 31B dense flagship across reasoning and vi As a third-party provider blog with deployment framing rather than a first-party Google announcement, the article is useful as independent corroboration of the 26B-A4B variant's active-parameter architecture and single-GPU deployability rather than as breaking news. It does not mention Cortecs and is oriented toward Si

According to a benchr.org launch record published July 28, 2026, Google released the instruction-tuned model with the exact identifier "gemma-4-26b-a4b-it" on April 2, 2026 as part of the Gemma 4 launch. The page records that this exact identifier is made available in AI Studio and via the Gemini API, and frames it as The same benchr record is explicit about what it does and does not publish: the checked release note does not include context-window limits, per-token API pricing, or benchmark results for the exact IT variant. It warns against filling those gaps from sibling Gemma models, older cards, or third-party listings, and trea

An independent deep-dive analysis published on kie.ai on July 16, 2026 examines Google's Gemma 4 family as a five-size open-weight lineup under Apache 2.0: E2B (2.3B effective), E4B (4.5B effective), 12B Unified (encoder-free multimodal), 26B-A4B MoE (25.2B total, 3.8B active per token), and 31B dense. The 12B, 26B-A4B The analysis reports that Official Arena AI text rankings place the 31B at Elo 1452 and the 26B-A4B at 1441, both above Gemma 3 27B's 1365, with the 31B scoring 80.0% on LiveCodeBench v6. Community NVFP4 quantizations from Unsloth reportedly deliver 1.5× speedup on NVIDIA Blackwell hardware, fitting the 12B into 11GB V

DeepInfra's analysis (May 25, 2026) characterizes Gemma 4 26B A4B as an open-weight Google DeepMind reasoning model with 25.2B total / 3.8B active MoE parameters, 262K context, native function calling, multimodal input, and Apache 2.0 licensing released April 3, 2026. It highlights blended provider pricing ranging from The piece positions DeepInfra on blended cost and TTFT, Clarifai on throughput and time-to-first-token, Cloudflare on output pricing and latency, and Google AI Studio on direct managed access, framing Gemma 4 as suited to cost-sensitive production, long-context assistants, and multimodal pipelines. It targets technical

Compare DeepSeek V4 Pro from DeepSeek and Gemma 4 26B A4B from Google on key metrics including benchmarks, price, context length, and other model features.

Gemma 4 26B A4B IT is an instruction-tuned Mixture-of-Experts (MoE) model from Google DeepMind. $0 per million input tokens, $0 per million output tokens. 262,144 token context window, maximum output of 32,768 tokens. Higher uptime with 12 providers. Includes independent benchmarks from Artificial Analysis.

NVIDIA's NGC catalog hosts Google Gemma 4 26B A4B IT packaged for NVIDIA NIM and marked supported by NVIDIA AI Enterprise, with signed BF16 and NVFP4 artifacts plus nim-assistant, nim-bf16 and nim-nvfp4 variants downloadable through the NGC CLI. The listing states the model is not owned or developed by NVIDIA and links to Google's own model card. The catalog describes gemma-4-26b-a4b-it as a Google-developed multimodal instruction-tuned Mixture-of-Experts model for text, image and video understanding, with roughly 25.2B total parameters and about 3.8B active parameters per token. It cites 256K-token long-context reasoning, coding, tool calling, multilingual and multi-turn conversation, and visual reasoning.

Google Cloud's Gemini Enterprise Agent Platform documentation lists Gemma 4 26B A4B IT as a multimodal Google model with model ID gemma-4-26b-a4b-it-maas, taking text and image input and producing text output. Function calling and structured output are supported, while audio, video, thinking, batch inference, and fixed-quota consumption are marked not supported. The same page reports a 128,000 maximum output and a 262,144 context length with a global quota, availability in the global multi-region and the us region, and provisioned throughput plus standard pay-as-you-go options. The variant is still at Experimental launch stage with a release date of April 3, 2026, and the page was last updated on 2026-10-16.

Google DeepMind publishes google/gemma-4-26B-A4B-it on Hugging Face under Apache 2.0, describing Gemma 4 as an open-weight multimodal family in five sizes (E2B, E4B, 12B, 26B A4B, 31B) with pre-trained and instruction-tuned variants. The card states Gemma 4 handles text and image input and generates text output, with audio native only on the E2B, E4B and 12B models. For the 26B A4B size specifically, the family supports a context window of up to 256K tokens and multilingual coverage in over 140 languages, and uses a hybrid attention mechanism interleaving local sliding-window attention with full global attention. Family-level gains include configurable thinking modes, native function calling for agentic workflows, and native system prompt support.

Third-party benchmark aggregator llm-stats.com ranks Gemma 4 26B-A4B at position 117 overall with a composite LLM Stats Score of 22.3 at a blended price of $0.14 per million tokens. The model's capability-tier placement is mixed: it lands in the top half for Data Viz (15 of 64), Audio (19 of 66), Websites (41 of 83), a Per-turn quality analysis shows the model declining from 14.2 at turn 1 to 13.6 at turns 31+, with the Quality Tracker showing +0.82σ improvement in the 7-day window (213 votes) driven by gains in Chat (+1.97σ), SVG (+2.02σ), and Games (+1.96σ), while declining in Websites (-0.64σ) and Audio (-0.63σ). Cost-efficiency c

The Kilo Code model page documents Gemma 4 26B A4B as available via OpenRouter with a 262,144-token context window, 16,384 max output tokens, and input modalities spanning text, image, and video. It confirms an April 3, 2026 creation date, the Gemma tokenizer, and pricing of $0.04 per 1M input tokens and $0.22 per 1M o The page lists coding-relevant capabilities including function calling, tool choice control, structured JSON schema outputs, and reasoning tokens support, while noting content moderation is disabled. A sample cost calculation shows analyzing a ~40k-token, 10k-token-output codebase at approximately $0.0039, giving devel

OpenRouter's official comparison page pits Google's Gemma 4 26B A4B IT against its larger sibling Gemma 4 31B IT, both sharing a 262K-token context window. Gemma 4 26B A4B is positioned as the more efficient MoE option at $0.07/M input and $0.34/M output, slightly cheaper than the 31B's $0.08/$0.35, with no cached-inpu On performance metrics, the 26B A4B variant delivers substantially better latency and throughput than its 31B counterpart, with a p50 latency of 0.34s versus 1.89s and p50 throughput of 23 tok/s versus 11 tok/s. Both support tool use, stream cancellation, and reasoning variants on the Artificial Analysis intelligence/a