$ diff gemini-3-6-flash longcat-2-0
Gemini 3.6 Flash vs LongCat-2.0
Values come from Gemini 3.6 Flash and LongCat-2.0, where each is cited to its source. This page states no benchmark result and ranks neither model — it puts two published specifications next to each other. Where a lab has not published a figure, the row says so rather than guessing.
Google DeepMindGemini 3.6 Flash
- context
- 1.0M
- weights
- closed
- $/M in
- $1.5
- $/M out
- $7.5
- context
- 1M
- weights
- open
- $/M in
- $0.3
- $/M out
- $1.2
What actually differs
- Context window
- Gemini 3.6 Flash takes 1.0M against 1M — modestly more room in a single request.
- Input price
- LongCat-2.0 at $0.3/M against $1.5/M — 5.0× cheaper to feed. Standard rates: one of these labs also quotes a lower cached-input tier, which applies only when a prefix is reused — see the full spec below.
- Output price
- LongCat-2.0 at $1.2/M against $7.5/M — 6.3× cheaper to generate. Output dominates the bill on most agentic workloads, where the model writes far more than it reads.
- Weights
- LongCat-2.0 publishes weights (MIT (commercially permissive)); Gemini 3.6 Flash is API-only. That decides self-hosting, air-gapped deployment and fine-tuning before any capability question does.
- Recency
- Gemini 3.6 Flash shipped 21 days after LongCat-2.0 (2026-07-21 vs 2026-06-30).
Full spec
| Attribute | Gemini 3.6 Flash | LongCat-2.0 |
|---|---|---|
| Developer | Google DeepMind | Meituan |
| Released | 2026-07-21 | 2026-06-30 |
| Context window | 1,048,576 tokens (1M) | 1 million tokens (LongCat Sparse Attention / LSA) |
| Pricing | $1.50/M input · $7.50/M output | $0.30/M input · $1.20/M output ($0.006/M cached input) — OpenRouter catalogue, read 2026-07-28. Open weights, so this is hosted-inference pricing rather than a licence fee. |
| License | proprietary | MIT (commercially permissive) |
| Availability | Gemini API, Google AI Studio, Android Studio, Vertex AI, consumer Gemini app, Google Search, GitHub Copilot | Open-sourced 2026-06-30; previously served on OpenRouter under the codename "Owl Alpha" |