← Model list
Sakana Namazu
sakana·sakana/sakana-namazu·GA·Closed·sakana-namazu series·NEW
Japanese-specialized reasoning model based on Kimi K2.6 and tuned for Japanese language, culture, and business workflows
At a glance
Official price$0.95 / $4.00 per 1M
Cache read $0.15 · Sakana AI
Lowest paid$0.95 / $4.00
OpenRouter Gateway · 1× spread
Context262,144
⚠ Providers report 256,000–262,144; the table below is authoritative
Output limit65,536
Capabilities
✓ Reasoning✓ Tool use✓ Structured output✓ Temperature✓ Attachments
Modalities
TextImagePDF
Knowledge cutoff—
Released / updated2026-08-03 / 2026-08-03
Quality & performance
Artificial Analysis doesn't cover this model (267 of 2059 have data). Quality data comes from independent evals covering widely used models.
Available at 5 providers5 with public prices
| Provider | Tier | Input | Output | Cache read | Cache write | Context | Output limit | Status |
|---|---|---|---|---|---|---|---|---|
| OpenRouter | Gateway | $0.95 | $4.00 | $0.15 | — | 262,144 | 65,536 | |
| Merge Gateway | Gateway | $0.95 | $4.00 | $0.15 | — | 262,144 | 65,536 | |
| Sakana AIOfficial | First-party | $0.95 | $4.00 | $0.15 | — | 262,144 | 65,536 | |
| Vercel AI Gateway sakana/namazu | Cloud | $0.95 | $4.00 | $0.15 | — | 256,000 ⚠ | 256,000 | |
| Kilo Gateway | Gateway | $0.95 | $4.00 | $0.15 | — | 262,144 | 65,536 |
Sorted by blended price (input×0.75 + output×0.25) asc. The official channel always shows regardless of rank. Whether a gateway's low price is actually usable can't be verified.
Your usage cost
1OpenRouter$294.00
2Merge Gateway$294.00
3Sakana AI · Official$294.00
4Vercel AI Gateway$294.00
5Kilo Gateway$294.00
Switch to OpenRouter to save $0.00/mo (0%). The gap is small, so staying on the official channel is fine.
Note: this is a gateway; verify availability and rate limits yourself.
Benchmark6 items
| Name | Conditions | Score | Metric | Source |
|---|---|---|---|---|
| AIME26 | — | 96.67 | — | Source ↗ |
| MMLU-Pro | — | 90.33 | — | Source ↗ |
| LiveCodeBench v6 | — | 90.33 | — | Source ↗ |
| JFBench | — | 37.4 | — | Source ↗ |
| Translation | — | 52.2 | — | Source ↗ |
| FairPoliticsQA | — | 56.3 | — | Source ↗ |
The same benchmark scores very differently across harness / dataset, so the qualifying conditions must be shown together.
Reasoning control
effort = noneeffort = highToggle (on / off)
2 / 5 providers expose no reasoning control (reasoning_options: []).
Related models
DeepSeek V4 Procheaper alternative$0.435 / $0.87DeepSeek V4 Flashcheaper alternative$0.14 / $0.28MiniMax-M2.7cheaper alternative$0.30 / $1.20DeepSeek V4 Flash 0731cheaper alternative$0.479 / $1.44
Price historyone sample accumulated per data sync
Input listOutput listMin blended