Kimi K2.8 Preview vs Macaron V1 Venti vs Sakana Namazu
Too close to call on our weighted score (Kimi K2.8 Preview 51, Sakana Namazu 50, Macaron V1 Venti 44). The right pick depends on what you value most.
Moonshot AIKimi K2.8 Preview
51/100- ECI—
- Price$1.00 / $4.00
- Context1.05M
Mindlab
Macaron V1 Venti
44/100- ECI—
- Price$0.825 / $2.48
- Context1.05M
Sakana AISakana Namazu
50/100- ECI—
- Price$0.95 / $4.00
- Context262K
Too close to call
It is close. Our weighted score puts them within 2 points (Kimi K2.8 Preview 51/100, Sakana Namazu 50/100, Macaron V1 Venti 44/100), so choose by what matters most for your work: Macaron V1 Venti on price. The score weighs price 50%, inputs & features 30%, context window 20%. None of the shared benchmarks cover every model here yet, so capability is left out of this verdict.
- CapabilityNot enough dataNo independent benchmark covers every model here yet
- Lowest priceMacaron V1 VentiMacaron V1 Venti $1.24 · Sakana Namazu $1.71 · Kimi K2.8 Preview $1.75 per 1M tokens (3:1 blend)
- Longest contextKimi K2.8 Preview and Macaron V1 VentiKimi K2.8 Preview 1,048,576 · Macaron V1 Venti 1,048,576 · Sakana Namazu 262,144 tokens
- Widest inputsKimi K2.8 Preview and Sakana NamazuKimi K2.8 Preview: Text, Images, Video · Macaron V1 Venti: Text · Sakana Namazu: Text, Images, PDFs
- Self-hostingMacaron V1 VentiPublishes downloadable weights (MIT)
| Measure | Weight | Kimi K2.8 Preview | Macaron V1 Venti | Sakana Namazu |
|---|---|---|---|---|
| Price | 50% | 37 | 44 | 37 |
| Inputs & features | 30% | 70 | 35 | 80 |
| Context window | 20% | 61 | 61 | 37 |
| Overall | 100% | 51/100 | 44/100 | 50/100 |
Left out because at least one model lacks the data: capability. The remaining weights were rescaled.
Every spec in one table
Highlighted cells lead their row. Dashes mean the data is not published.
| Specification | Kimi K2.8 PreviewMoonshot AI | Macaron V1 VentiMindlab | Sakana NamazuSakana AI |
|---|---|---|---|
| Capability | |||
| Capabilities Index (ECI) | — | — | — |
| ECI rank | — | — | — |
| Price per million tokens | |||
| Input | $1.00 | $0.825 (best) | $0.95 |
| Output | $4.00 | $2.48 (best) | $4.00 |
| Cached input | — | — | $0.15 |
| Blended (3:1) | $1.75 | $1.24 (best) | $1.71 |
| Long-context rate | Same rate | Same rate | Same rate |
| Price source | Median of 2 providers | Median of 1 providers | Official Sakana AI API |
| Limits | |||
| Context window | 1,048,576 tokens (best) | 1,048,576 tokens (best) | 262,144 tokens |
| Max output | — | 131,072 tokens (best) | 65,536 tokens |
| Inputs and features | |||
| Text | Yes | Yes | Yes |
| Images | Yes | No | Yes |
| PDFs | No | No | Yes |
| Audio | No | No | No |
| Video | Yes | No | No |
| Reasoning | Yes | Yes | Yes |
| Tool calling | Yes | Yes | Yes |
| Structured output | No | No | Yes |
| Availability | |||
| Weights | Proprietary | OpenMIT | Proprietary |
| API model ID | — | — | sakana-namazu |
| API providers | 2 | 1 | 5 (best) |
| Released | Sep 11, 2026 | Jul 21, 2026 | Aug 3, 2026 |
| Knowledge cutoff | — | — | — |
What would a month cost?
Enter your expected volume in millions of tokens. List prices only; caching and batch discounts would lower these.
- Kimi K2.8 Preview$18.00
- Macaron V1 Venti$13.20
- Sakana Namazu$17.50
Which should you choose?
Which is better: Kimi K2.8 Preview, Macaron V1 Venti or Sakana Namazu?
It is close. Our weighted score puts them within 2 points (Kimi K2.8 Preview 51/100, Sakana Namazu 50/100, Macaron V1 Venti 44/100), so choose by what matters most for your work: Macaron V1 Venti on price. The score weighs price 50%, inputs & features 30%, context window 20%. None of the shared benchmarks cover every model here yet, so capability is left out of this verdict.
Which is cheaper, Kimi K2.8 Preview, Macaron V1 Venti or Sakana Namazu?
Macaron V1 Venti is cheaper at $0.825 input / $2.48 output per million tokens (median across 1 API provider). Sakana Namazu costs $0.95 input / $4.00 output per million tokens (official Sakana AI API price); Kimi K2.8 Preview costs $1.00 input / $4.00 output per million tokens (median across 2 API providers). At a typical mix of three input tokens to one output token, that is $1.24 per million tokens for Macaron V1 Venti versus $1.71 for Sakana Namazu (1.4× as much) and $1.75 for Kimi K2.8 Preview (1.4× as much).
Which scores higher on benchmarks?
There is no independent benchmark that covers all three models yet. Kimi K2.8 Preview has not been scored yet, Macaron V1 Venti has not been scored yet and Sakana Namazu has not been scored yet.
Which is better for coding?
There are no published SWE-bench Verified results for Kimi K2.8 Preview, Macaron V1 Venti and Sakana Namazu yet, so there is no like-for-like coding score. Test both on a sample of your own tasks. All three support tool calling for agent workflows.
Which has the bigger context window?
Kimi K2.8 Preview and Macaron V1 Venti have the largest context windows (1,048,576 and 1,048,576 tokens), against 262,144 for Sakana Namazu. Maximum output per response: Macaron V1 Venti up to 131,072, Sakana Namazu up to 65,536 tokens.
Which can read images, PDFs, audio or video?
Kimi K2.8 Preview accepts text, images and video; Macaron V1 Venti accepts text; Sakana Namazu accepts text, images and PDFs. Kimi K2.8 Preview handles the widest range of inputs.
Are any of these open source?
Macaron V1 Venti publishes its weights (MIT) and can be self-hosted; Kimi K2.8 Preview and Sakana Namazu is proprietary.
Which is newer?
Kimi K2.8 Preview is the newest, released Sep 11, 2026. Sakana Namazu came out Aug 3, 2026; Macaron V1 Venti came out Jul 21, 2026.
How do you decide the winner?
Each model gets a 0–100 score on capability (50%, independent benchmark results); price (25%, blended price per million tokens (3 input : 1 output), log scale); inputs & features (15%, image, PDF, audio and video input, tool calling, structured output and reasoning); context window (10%, maximum tokens per request, log scale). Dimensions missing for any model are dropped and the remaining weights rescaled, so every model is judged on the same evidence. Specs and prices come from public model listings and the labs’ own API pages; capability scores come from independent benchmark runs. Data updated Nov 2, 2026.
