Yi Model vs Qwen vs DeepSeek vs Llama
A side-by-side look at scores, pricing and features — with RECATOOLS' ASEAN-aware verdict for each.
Yi Model
01.AI's open bilingual models, frozen since 2024; the paid API closed...
Visit
|
Qwen
Alibaba Cloud's LLM family, open weights through to the Qwen3.8-Max fl...
Visit
|
DeepSeek
Hangzhou lab shipping MIT-licensed V4 models and a very cheap API
Visit
|
Llama
Meta's downloadable LLM family, last updated with Llama 4 in April 202...
Visit
|
|
|---|---|---|---|---|
| RECATOOLS Score | 6.8 / 10 | 8.6 / 10 | 8.2 / 10 | 8.3 / 10 |
| Capability | ||||
| Value for money | ||||
| Ease of use | ||||
| ASEAN readiness | ||||
| API quality | ||||
| Pricing | Open Source | Freemium | Open Source | Open Source |
| Free tier | Yi-6B, Yi-34B, Yi-1.5 and Yi-Coder weights free under Apache 2.0; the paid Yi API closed on 3 September 2026 | Open weights to self-host; free Qwen Chat; 1M free tokens per model for 90 days in Model Studio's Singapore region | Free chat on the web and in the app; MIT-licensed open weights to self-host | Free weight downloads (Llama 2, 3.x and 4) under Meta's Llama community licences, after accepting the licence |
| Paid from | — | From $0.03 per million input tokens (qwen3.7-flash, prompts up to 32K, Singapore region) | $0.15 per 1M input tokens (deepseek-flash, cache miss, off-peak); $0.30 at peak | — |
| Has API | ✓ | ✓ | ✓ | ✓ |
| Open source | ✓ | ✓ | ✓ | ✓ |
| Free to use | ✓ | ✓ | ✓ | ✓ |
| Users | 500k+ downloads | Consumer app ~203M MAU (April 2026), #3 globally | ~130M monthly active users (late 2025) | 10m+ downloads |
| Founded | 2023 | 2023 | 2023 | 2023 |
| Maker | 01.AI (零一万物) | Alibaba Cloud | High-Flyer | Meta Platforms |
| Verdict | Yi is now a finished product line. The open weights, from Yi-6B to Yi-1.5 and Yi-Coder, remain free under Apache 2.0 and are still a usable bilingual Chinese-English base, especially for teams that already deployed them.... |
Qwen is the most practical open-weight family to build on in 2026: Apache 2.0 models from under a billion to hundreds of billions of parameters, and now even the Max-tier flagship available to download. Hosted pricing in... |
DeepSeek remains the price benchmark for capable models: V4.1-Flash costs $0.15 to $0.30 per million input tokens and V4-Pro under $4 per million output at peak, both with 1M-token context, and the weights are MIT-licens... |
Llama 2 and 3 made Meta's family the default starting point for self-hosted and fine-tuned models, and the weights from Llama 2 through Llama 4 remain free to download from Hugging Face and Kaggle. Llama 4 Scout and Mave... |
| Full review → | Full review → | Full review → | Full review → |
Comparisons cover up to 4 tools. Scores are RECATOOLS editorial assessments; verify current pricing on each vendor's site.