Sakana: Fugu Max vs Qwen: Qwen3.8 27B
Side-by-side comparison of specs, pricing, benchmark scores, and task rankings. Updated 2026-10-04.
| Sakana: Fugu Max | Qwen: Qwen3.8 27B | |
|---|---|---|
| Vendor | sakana | qwen |
| Quality Score | 100 | 100 |
| Benchmark Score | - | 55.6 |
| Input Price | $2.00/M | $0.42/M |
| Output Price | $6.00/M | $3.00/M |
| Context Window | 1,000,000 | 1,000,000 |
| Max Output | 128,000 | 131,072 |
| Tool Calling | ✓ | ✓ |
| Structured Output | ✓ | ✓ |
| Reasoning Mode | ✓ | ✓ |
| Vision | ✓ | ✓ |
| Audio | - | - |
| Benchmark Scores | ||
| ai_index | - | 55.6 |
The verdict
For a sample job of 1 million input tokens and 250,000 output tokens, Qwen3.8 27B costs $1.17 and Fugu Max costs $3.50, so Qwen3.8 27B is about 3.0 times cheaper for the same work.
Both read up to 1,000,000 tokens at once, roughly 1,500 pages of text.
Only Qwen3.8 27B offers video input.
Across the 20 tasks where they differ, Qwen3.8 27B on 20 (SQL Generation, Code Review, Code Completion and others).
Fugu Max was released in September 2026 and Qwen3.8 27B in August 2026.
Pick Qwen3.8 27B if you need lower cost, video input or the stronger all-round task scores.
Costs use current list prices per million tokens; page counts assume about 0.75 words per token and 500 words per page.
Who wins by task?
| Task | Sakana: Fugu Max | Qwen: Qwen3.8 27B |
|---|---|---|
| SQL Generation | 133 | 140 |
| Code Review | 132 | 142 |
| Code Completion | 118 | 132 |
| Code Refactoring | 136 | 144 |
| Bug Fixing | 136 | 146 |
| Unit Test Generation | 124 | 130 |
| Code Documentation | 130 | 136 |
| Regex Writing | 118 | 124 |
| CI/CD Pipelines | 120 | 126 |
| Frontend Component Design | 122 | 128 |
| Data Analysis | 124 | 132 |
| CSV / Spreadsheet Cleanup | 133 | 135 |
| ETL Scripting | 128 | 135 |
| JSON Extraction | 122 | 131 |
| Bulk Data Labeling | 119 | 129 |
| OCR / Document Parsing | 131 | 133 |
| Table Extraction from PDFs | 131 | 133 |
| Long-Document Summarization | 137 | 146 |
| Short-Form Summarization | 114 | 126 |
| Blog Post Writing | 121 | 128 |
Scores reflect capability match + benchmark data + pricing for each task. Methodology →