DeepSeek-R1 vs Kimi K2 Instruct vs Qwen3 VL Flash (Alibaba Cloud)
Qwen3 VL Flash (Alibaba Cloud) comes out ahead, 72 to 36 and 34 on our weighted score, and it is the cheaper option too.
DeepSeekDeepSeek-R1
34/100- ECI138.8
- Price$0.70 / $4.88
- Context128K
Moonshot AIKimi K2 Instruct
36/100- ECI139.9
- Price$0.57 / $2.30
- Context131K
- Our pick
Alibaba (Qwen)Qwen3 VL Flash (Alibaba Cloud)
72/100- ECI—
- Price$0.022 / $0.215
- Context262K
Qwen3 VL Flash (Alibaba Cloud) is our pick
Qwen3 VL Flash (Alibaba Cloud) is the better all-round choice, scoring 72/100 against Kimi K2 Instruct (36) and DeepSeek-R1 (34). It leads on price, inputs & features and context window. 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 priceQwen3 VL Flash (Alibaba Cloud)Qwen3 VL Flash (Alibaba Cloud) $0.07 · Kimi K2 Instruct $1.00 · DeepSeek-R1 $1.75 per 1M tokens (3:1 blend)
- Longest contextQwen3 VL Flash (Alibaba Cloud)Qwen3 VL Flash (Alibaba Cloud) 262,144 · Kimi K2 Instruct 131,072 · DeepSeek-R1 128,000 tokens
- Widest inputsQwen3 VL Flash (Alibaba Cloud)DeepSeek-R1: Text · Kimi K2 Instruct: Text · Qwen3 VL Flash (Alibaba Cloud): Text, Images
- Self-hostingDeepSeek-R1 and Kimi K2 InstructPublishes downloadable weights (Modified MIT)
| Measure | Weight | DeepSeek-R1 | Kimi K2 Instruct | Qwen3 VL Flash (Alibaba Cloud) |
|---|---|---|---|---|
| Price | 50% | 37 | 48 | 100 |
| Inputs & features | 30% | 35 | 25 | 50 |
| Context window | 20% | 24 | 25 | 37 |
| Overall | 100% | 34/100 | 36/100 | 72/100 |
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 | DeepSeek-R1DeepSeek | Kimi K2 InstructMoonshot AI | Qwen3 VL Flash (Alibaba Cloud)Alibaba (Qwen) |
|---|---|---|---|
| Capability | |||
| Capabilities Index (ECI) | 138.8 | 139.9 (best) | — |
| ECI rank | #119 of 165 | #112 of 165 (best) | — |
| GPQA DiamondGraduate-level science questions | 71.7% | — | — |
| OTIS Mock AIME 2024–2025Competition mathematics | 53.3% | — | — |
| Price per million tokens | |||
| Input | $0.70 | $0.57 | $0.022 (best) |
| Output | $4.88 | $2.30 | $0.215 (best) |
| Cached input | — | — | — |
| Blended (3:1) | $1.75 | $1.00 | $0.07 (best) |
| Long-context rate | Same rate | Same rate | Same rate |
| Price source | Median of 11 providers | Median of 11 providers | Median of 1 providers |
| Limits | |||
| Context window | 128,000 tokens | 131,072 tokens | 262,144 tokens (best) |
| Max output | 32,768 tokens | 32,768 tokens | 32,768 tokens |
| Inputs and features | |||
| Text | Yes | Yes | Yes |
| Images | No | No | Yes |
| PDFs | No | No | No |
| Audio | No | No | No |
| Video | No | No | No |
| Reasoning | Yes | No | No |
| Tool calling | Yes | Yes | Yes |
| Structured output | No | No | No |
| Availability | |||
| Weights | Open | OpenModified MIT | Proprietary |
| API model ID | — | — | — |
| API providers | 12 (best) | 11 | 1 |
| Released | Jan 20, 2025 | Jul 11, 2025 | Oct 15, 2025 |
| Knowledge cutoff | Jul 2024 | — | — |
What would a month cost?
Enter your expected volume in millions of tokens. List prices only; caching and batch discounts would lower these.
- DeepSeek-R1$16.76
- Kimi K2 Instruct$10.30
- Qwen3 VL Flash (Alibaba Cloud)$0.65
Which should you choose?
Which is better: DeepSeek-R1, Kimi K2 Instruct or Qwen3 VL Flash (Alibaba Cloud)?
Qwen3 VL Flash (Alibaba Cloud) is the better all-round choice, scoring 72/100 against Kimi K2 Instruct (36) and DeepSeek-R1 (34). It leads on price, inputs & features and context window. 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, DeepSeek-R1, Kimi K2 Instruct or Qwen3 VL Flash (Alibaba Cloud)?
Qwen3 VL Flash (Alibaba Cloud) is cheaper at $0.022 input / $0.215 output per million tokens (median across 1 API provider). Kimi K2 Instruct costs $0.57 input / $2.30 output per million tokens (median across 11 API providers); DeepSeek-R1 costs $0.70 input / $4.88 output per million tokens (median across 11 API providers). At a typical mix of three input tokens to one output token, that is $0.07 per million tokens for Qwen3 VL Flash (Alibaba Cloud) versus $1.00 for Kimi K2 Instruct (14× as much) and $1.75 for DeepSeek-R1 (25× as much).
Which scores higher on benchmarks?
There is no independent benchmark that covers all three models yet. DeepSeek-R1 has an ECI of 138.8, Kimi K2 Instruct has an ECI of 139.9 and Qwen3 VL Flash (Alibaba Cloud) has not been scored yet.
Which is better for coding?
There are no published SWE-bench Verified results for DeepSeek-R1, Kimi K2 Instruct and Qwen3 VL Flash (Alibaba Cloud) 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?
Qwen3 VL Flash (Alibaba Cloud) has the largest context window at 262,144 tokens, against 131,072 for Kimi K2 Instruct and 128,000 for DeepSeek-R1. Maximum output per response: DeepSeek-R1 up to 32,768, Kimi K2 Instruct up to 32,768, Qwen3 VL Flash (Alibaba Cloud) up to 32,768 tokens.
Which can read images, PDFs, audio or video?
DeepSeek-R1 accepts text; Kimi K2 Instruct accepts text; Qwen3 VL Flash (Alibaba Cloud) accepts text and images. Qwen3 VL Flash (Alibaba Cloud) handles the widest range of inputs.
Are any of these open source?
DeepSeek-R1 and Kimi K2 Instruct publishes its weights (Modified MIT) and can be self-hosted; Qwen3 VL Flash (Alibaba Cloud) is proprietary.
Which is newer?
Qwen3 VL Flash (Alibaba Cloud) is the newest, released Oct 15, 2025. Kimi K2 Instruct came out Jul 11, 2025; DeepSeek-R1 came out Jan 20, 2025. Knowledge cutoff: DeepSeek-R1 Jul 2024.
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.
Related comparisons
- DeepSeek-R1 vs GPT-4.1OpenAI · ECI 136.7
- DeepSeek-R1 vs QwQ 32BAlibaba (Qwen) · ECI 137.5
- DeepSeek-R1 vs Gemini 2.5 FlashGoogle · ECI 140.6
- Kimi K2 Instruct vs Gemini 2.5 FlashGoogle · ECI 140.6
- Kimi K2 Instruct vs DeepSeek R1 0528DeepSeek · ECI 141.2
- Kimi K2 Instruct vs Qwen3 235B-A22B Thinking 2507Alibaba (Qwen) · ECI 143.8
