Gemma 4 26B A4B
Gemma 4 26B A4B by Google: 15 offers. From €0.063 per 1M input tokens, with EU processing from €0.10.
Calculate costs for this modelCompare offersView price history
Profile
| Profile | |
|---|---|
| Developer | |
| Type | Language models |
| Origin | US |
| Input types | text |
| Context (tokens) | 256,000 |
| Max. output (tokens) | 32,768 |
| Reasoning | ohne |
| Parameters (bn) | 26 |
| Open weights | ✓ yes |
| Licence | Apache-2.0 |
| VRAM at 4-bit (GB) | 19 (RTX PRO 4000) |
| Released | – |
searchit Entity Extraction (24 Sept 2026): 25.2 %
Offers with processing in the EU (5)
| Provider | Country | Variant | Input | Output | Source | As of |
|---|---|---|---|---|---|---|
| AKI.IO | DE | standard | €0.10 | €0.50 | aki.io | 13 Oct 2026 |
| cortecs | AT | standard | €0.10 | €0.50 | api.cortecs.ai | 14 Oct 2026 |
| evroc | SE | standard | €0.12 | €0.50 | evroc.com | 8 Oct 2026 |
| HostYourAI | NL | standard | €0.12 | €0.58 | hostyourai.com | 14 Oct 2026 |
| Scaleway | FR | standard | €0.25 | €0.50 | www.scaleway.com | 13 Oct 2026 |
€ per 1M tokens. Source: providers' pricing pages, read and checked by searchit. As of 14 Oct 2026.
Offers outside the EU (10)
| Provider | Country | Variant | Input | Output | Source | As of |
|---|---|---|---|---|---|---|
| Cloudflare Workers AI | US | standard | €0.089 | €0.27 | developers.cloudflare.com | 10 Oct 2026 |
| DeepInfra | US | standard | €0.063 | €0.30 | deepinfra.com | 14 Oct 2026 |
| io.net | US | standard | €0.088 | €0.29 | api.intelligence.io.solutions | 14 Oct 2026 |
| Novita AI | US | standard | €0.12 | €0.36 | novita.ai | 13 Oct 2026 |
| Parasail | US | standard | €0.12 | €0.36 | www.parasail.io | 2 Oct 2026 |
| Parasail | US | serverless | €0.12 | €0.36 | www.parasail.io | 2 Oct 2026 |
| Venice.ai | US | standard | €0.12 | €0.36 | docs.venice.ai | 14 Oct 2026 |
| SiliconFlow | SG | standard | €0.13 | €0.36 | www.siliconflow.com | 2 Oct 2026 |
| Google Vertex AI | US | standard | €0.13 | €0.54 | cloud.google.com | 14 Oct 2026 |
| Modular (Modular Cloud) | US | standard | €0.13 | €0.54 | www.modular.com | 14 Oct 2026 |
€ per 1M tokens. Source: providers' pricing pages, read and checked by searchit. As of 14 Oct 2026.
AI Act
Whether the developer has signed the Code of Practice for general-purpose AI models: unknown. What this means is explained under “EU and AI Act”. EU processing and AI Act
Benchmarks
| Benchmarks | |
|---|---|
| GPQA Diamond (Artificial Analysis) | 79.2 |
| Humanity's Last Exam (Artificial Analysis) | 19.3 |
| OTIS Mock AIME 2024-2025 (Epoch AI) | 82.2 |
| ExtractBench: A Benchmark for Schema-Guided Enterprise Document Extraction (arXiv:2607.29677) | 66.2 |
| ExtractBench: A Benchmark for Schema-Guided Enterprise Document Extraction (arXiv:2607.29677) – value f1 gesamt | 66.2 |
| ExtractBench: A Benchmark for Schema-Guided Enterprise Document Extraction (arXiv:2607.29677) – value f1 kurz | 80.5 |
| ExtractBench: A Benchmark for Schema-Guided Enterprise Document Extraction (arXiv:2607.29677) – value f1 lang | 12.2 |
| ExtractBench: A Benchmark for Schema-Guided Enterprise Document Extraction (arXiv:2607.29677) – value f1 mittel | 40.5 |
| ExtractBench: A Benchmark for Schema-Guided Enterprise Document Extraction (arXiv:2607.29677) – value precision gesamt | 67 |
| ExtractBench: A Benchmark for Schema-Guided Enterprise Document Extraction (arXiv:2607.29677) – value recall gesamt | 66 |
| Vectara HHEM-2.3 | 5.2 |
| Vectara HHEM-2.3 – antwortrate | 99.8 |
| Vectara HHEM-2.3 – halluzination | 5.2 |
| MMMU-Pro (llm-stats-Abzug, Rang 1..50 von 69) | 0.74 |
| searchit Extraktions-Messlauf REPO-9 (PROMPT-CRAWLER-TEXT-START, 8 Dokumente × 3 Läufe) – treffer beziehungen | 11.7 |
| searchit Extraktions-Messlauf REPO-9 (PROMPT-CRAWLER-TEXT-START, 8 Dokumente × 3 Läufe) – treffer entitaeten | 38.7 |