DeepSeek VL2 Small vs Qwen2-VL-72B-Instruct

    Comparison

    As of August 30, 2026, DeepSeek VL2 Small is 6% cheaper per million tokens than Qwen2-VL-72B-Instruct. DeepSeek VL2 Small is $0.20 / $1.50 per million input/output tokens versus Qwen2-VL-72B-Instruct at $0.80 / $1.00. Sources: https://api-docs.deepseek.com/quick_start/pricing and https://openrouter.ai/qwen/qwen2.5-vl-72b-instruct

    Sources: DeepSeek API pricing (https://api-docs.deepseek.com/quick_start/pricing); OpenRouter pricing (https://openrouter.ai/qwen/qwen2.5-vl-72b-instruct)

    Better value

    DeepSeek VL2 Small

    DeepSeek

    $1.70blended / 1M

    Input

    $0.20

    Output

    $1.50

    129.3K ctx|Open Source
    6%
    cheaper

    Qwen2-VL-72B-Instruct

    Qwen

    $1.80blended / 1M

    Input

    $0.80

    Output

    $1.00

    131.1K ctx|Open Source
    Save $0.10 per million tokens by choosing DeepSeek VL2 Small over Qwen2-VL-72B-Instruct.

    DeepSeek VL2 Small vs Qwen2-VL-72B-Instruct comparison

    Token pricing, specs, and benchmarks side by side

    Shared benchmarks

    Each bar is share of the current leader on that eval. Click a row to open the board.

    Score vs price

    Chartqa. Left is cheaper. Up is a higher score. The line is the best score you can buy at each price.

    Best score at each priceDeepSeek VL2 SmallQwen2-VL-72B-Instruct
    Metric
    DeepSeek VL2 Small
    Qwen2-VL-72B-Instruct
    Provider
    Provider
    DeepSeek
    Qwen
    License
    Open Source
    Open Source
    Release Date
    2024-12-13
    2024-08-29
    Pricing (per 1M tokens)
    Input Price-75%
    $0.20
    $0.80
    Output Price+50%
    $1.50
    $1.00
    Blended (1M + 1M)
    $1.70
    $1.80
    Model Details
    Context Window
    129.3K
    131.1K
    Max Output Tokens
    N/A
    N/A
    Knowledge Cutoff
    N/A
    2023-06-30
    Throughput
    N/A
    N/A
    Benchmarks
    46.2%
    48%
    80%
    84.5%
    88.3%
    92.3%
    96.5%
    77.9%
    75.8%
    84.5%
    60.7%
    70.5%
    80.3%
    86.5%
    79.3%
    21.2%
    64.5%
    57%
    62.9%
    30.9%
    73.6%
    83.4%
    87.7%
    65.4%
    77.8%
    83.4%
    85.5%
    91.9%
    71.2%
    Benchmark Wins
    DeepSeek VL2 Small 0|6 Qwen2-VL-72B-Instruct

    Verdict

    DeepSeek VL2 Small vs Qwen2-VL-72B-Instruct: the bottom line

    DeepSeek VL2 Small offers significantly lower pricing, while Qwen2-VL-72B-Instruct leads on benchmark performance. Your choice depends on whether cost efficiency or raw capability matters more for your use case.

    DeepSeek VL2 Small vs Qwen2-VL-72B-Instruct FAQ

    Which is cheaper, DeepSeek VL2 Small or Qwen2-VL-72B-Instruct?

    As of August 30, 2026, DeepSeek VL2 Small is 6% cheaper on blended LLM API pricing. DeepSeek VL2 Small costs $0.20 per million input tokens and $1.50 per million output tokens ($1.70 blended 1M-in + 1M-out). Qwen2-VL-72B-Instruct costs $0.80 / $1.00 per million tokens ($1.80 blended). Sources: https://api-docs.deepseek.com/quick_start/pricing and https://openrouter.ai/qwen/qwen2.5-vl-72b-instruct. The cheapest LLM for your app still depends on how many output tokens you generate.

    How much does DeepSeek VL2 Small vs Qwen2-VL-72B-Instruct cost per million tokens?

    DeepSeek VL2 Small DeepSeek API pricing is $0.20 input and $1.50 output per million tokens via Replicate. Qwen2-VL-72B-Instruct OpenRouter API pricing is $0.80 input and $1.00 output per million tokens via OpenRouter. Use those four numbers, not ChatGPT Plus or Claude Pro subscription prices, when you are comparing APIs.

    Which is better for coding, DeepSeek VL2 Small or Qwen2-VL-72B-Instruct?

    Neither DeepSeek VL2 Small nor Qwen2-VL-72B-Instruct has a published SWE-bench or LiveCodeBench score on this page yet. Use the comparison table for the evals that do exist.

    What is the context window for DeepSeek VL2 Small vs Qwen2-VL-72B-Instruct?

    DeepSeek VL2 Small supports a 129K token context window, while Qwen2-VL-72B-Instruct supports 131K tokens. Qwen2-VL-72B-Instruct offers a larger context window. Filling a larger window bills more input tokens, so the cheaper-per-million model can still cost more on long documents.

    Which model performs better on benchmarks, DeepSeek VL2 Small or Qwen2-VL-72B-Instruct?

    Qwen2-VL-72B-Instruct leads on 6 of 6 shared benchmarks versus DeepSeek VL2 Small's 0 wins. Check the comparison table for GPQA Diamond, SWE-bench, MMLU, HLE, and the other evals we track. Token price and benchmark score together are the usual LLM comparison, not either number alone.

    Is DeepSeek VL2 Small or Qwen2-VL-72B-Instruct better for production use?

    Both DeepSeek VL2 Small and Qwen2-VL-72B-Instruct are production API models. For cost-sensitive production traffic, DeepSeek VL2 Small has better economics on the blended 1M-in + 1M-out scale. For maximum capability, weight the benchmark table for your domain. Many production stacks route cheap models for drafts and a frontier model for the hard turn, which is usually cheaper than sending everything to the expensive API.

    Can I switch between DeepSeek VL2 Small and Qwen2-VL-72B-Instruct in my app?

    Yes. Call DeepSeek VL2 Small through Replicate and Qwen2-VL-72B-Instruct through OpenRouter with separate API keys, or through a gateway that already wraps both. Keep prompts in tokens, not characters, when you estimate the invoice. AnotherWrapper templates can swap providers without rewriting auth and billing.

    How accurate is this DeepSeek VL2 Small vs Qwen2-VL-72B-Instruct pricing data?

    List rates are the latest published DeepSeek API pricing and OpenRouter API pricing, quoted per million tokens. Official rate sheets: https://api-docs.deepseek.com/quick_start/pricing and https://openrouter.ai/qwen/qwen2.5-vl-72b-instruct. Prompt caching, batch APIs, and committed-use discounts can change the invoice. Benchmark scores are from official publications and independent evals, updated as new numbers land.

    What is the output token limit for DeepSeek VL2 Small and Qwen2-VL-72B-Instruct?

    DeepSeek VL2 Small supports up to an unspecified number of output tokens per request, while Qwen2-VL-72B-Instruct supports up to an unspecified number of output tokens. Output tokens are usually the expensive half of API pricing, so a higher max output cap is a capability, not a discount.

    Which model has better throughput, DeepSeek VL2 Small or Qwen2-VL-72B-Instruct?

    Neither DeepSeek VL2 Small nor Qwen2-VL-72B-Instruct has published throughput data on this page yet. Throughput (tokens per second) and time-to-first-token are why two models with the same per-million-token price can still feel different in chat and agents.

    How do I switch the models in this DeepSeek VL2 Small vs Qwen2-VL-72B-Instruct comparison?

    Use the model pickers at the top of this LLM comparison. Search any model in the index and the URL updates. Open DeepSeek VL2 Small or Qwen2-VL-72B-Instruct from the sidebar for the single-model API pricing page, or go back to the LLM pricing table to sort by cheapest blended cost.

    DeepSeek VL2 Small vs Qwen2-VL-72B-Instruct pricing

    This comparison covers API pricing, context window, throughput, and benchmark scores. DeepSeek VL2 Small costs $0.20/M input and $1.50/M output via Replicate. Qwen2-VL-72B-Instruct costs $0.80/M input and $1.00/M output via OpenRouter.

    Rank both on the LLM leaderboard, SWE-bench, and Humanity's Last Exam.

    The index

    All Large Language Models

    343 models across 36 providers. Search or jump to a lab — every model page stays linked here.

    Baidu

    2 models

    Inception

    1 models

    inclusionAI

    1 models

    LG AI Research

    1 models

    Liquid AI

    2 models

    Nous Research

    1 models

    OpenBMB

    1 models

    Sakana AI

    1 models

    Sarvam AI

    2 models

    Tencent

    1 models

    Thinking Machines

    1 models

    Unisound

    1 models

    Upstage

    1 models

    Next step

    You found the model. Now ship the product.

    Auth, billing, and the API layer are already decided. Fork 8 finished AI apps, or have us build the first version with you.