ZMIME
Comparison · 3 models · Updated Nov 18, 2026

Qwen3.5 397B-A17B vs Grok 4.20 (Non-Reasoning) vs Kimi K2.5

No independent capability benchmark covers every model here yet. The score uses price, inputs & features and context window.

Price in rating: Included · Exclude price

  1. Alibaba (Qwen)

    Qwen3.5 397B-A17B

    Released Feb 15, 2026

    54/100
    • ECI148.0
    • Price$0.60 / $3.60
    • Context262K
  2. xAI

    Grok 4.20 (Non-Reasoning)

    Released Mar 9, 2026

    53/100
    • ECI—
    • Price$1.25 / $2.50
    • Context1M
  3. Moonshot AI

    Kimi K2.5

    Released Jan 27, 2026

    52/100
    • ECI148.8
    • Price$0.60 / $3.00
    • Context262K
01 — Verdict

Capability evidence is incomplete

There is not enough shared capability evidence to name an overall winner. The scores (Qwen3.5 397B-A17B 54/100, Grok 4.20 (Non-Reasoning) 53/100, Kimi K2.5 52/100) compare the available specifications and 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 priceKimi K2.5Kimi K2.5 $1.20 · Qwen3.5 397B-A17B $1.35 · Grok 4.20 (Non-Reasoning) $1.56 per 1M tokens (3:1 blend)
  • Longest contextGrok 4.20 (Non-Reasoning)Grok 4.20 (Non-Reasoning) 1,000,000 · Qwen3.5 397B-A17B 262,144 · Kimi K2.5 262,144 tokens
  • Widest inputsSame inputsQwen3.5 397B-A17B: Text, Images, Video · Grok 4.20 (Non-Reasoning): Text, Images, PDFs · Kimi K2.5: Text, Images, Video
  • Self-hostingQwen3.5 397B-A17B and Kimi K2.5Publishes downloadable weights
How the score is built
MeasureWeightQwen3.5 397B-A17BGrok 4.20 (Non-Reasoning)Kimi K2.5
Price50%423944
Inputs & features30%807070
Context window20%436443
Specs & price only100%54/10053/10052/100

Left out because at least one model lacks the data: capability. The remaining weights were rescaled.

02 — Side by side

Every spec in one table

Highlighted cells lead their row. Dashes mean the data is not published.

Qwen3.5 397B-A17B vs Grok 4.20 (Non-Reasoning) vs Kimi K2.5 specifications side by side
SpecificationQwen3.5 397B-A17BAlibaba (Qwen)Grok 4.20 (Non-Reasoning)xAIKimi K2.5Moonshot AI
Capability
Capabilities Index (ECI)148.0—148.8 (best)
ECI rank#76 of 158—#73 of 158 (best)
GPQA DiamondGraduate-level science questions86.4%—87.6% (best)
FrontierMath Tiers 1–3Research-level mathematics31.2%——
OTIS Mock AIME 2024–2025Competition mathematics88.9%—92.2% (best)
SWE-bench VerifiedFixing real GitHub issues——73.8%
SimpleQA VerifiedShort factual questions——34.3%
Price per million tokens
Input$0.60 (best)$1.25$0.60 (best)
Output$3.60$2.50 (best)$3.00
Cached input—$0.20—
Blended (3:1)$1.35$1.56$1.20 (best)
Long-context rateSame rateOver 200K: $2.50 / $5.00Same rate
Price sourceOfficial Alibaba APIOfficial xAI APIMedian of 36 providers
Limits
Context window262,144 tokens1,000,000 tokens (best)262,144 tokens
Max output65,536 tokens30,000 tokens262,144 tokens (best)
Inputs and features
TextYesYesYes
ImagesYesYesYes
PDFsNoYesNo
AudioNoNoNo
VideoYesNoYes
ReasoningYesNoYes
Tool callingYesYesYes
Structured outputYesYesNo
Availability
WeightsOpenProprietaryOpen
API model IDqwen3.5-397b-a17bgrok-4.20-0309-non-reasoning—
API providers36 (best)1336 (best)
ReleasedFeb 15, 2026Mar 9, 2026Jan 27, 2026
Knowledge cutoff——Jan 2025
03 — Cost

What would a month cost?

Enter your expected volume in millions of tokens. List prices only; caching and batch discounts would lower these.

  • Qwen3.5 397B-A17B$13.20
  • Grok 4.20 (Non-Reasoning)$17.50
  • Kimi K2.5$12.00
04 — Questions

Which should you choose?

Which is better: Qwen3.5 397B-A17B, Grok 4.20 (Non-Reasoning) or Kimi K2.5?

There is not enough shared capability evidence to name an overall winner. The scores (Qwen3.5 397B-A17B 54/100, Grok 4.20 (Non-Reasoning) 53/100, Kimi K2.5 52/100) compare the available specifications and 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, Qwen3.5 397B-A17B, Grok 4.20 (Non-Reasoning) or Kimi K2.5?

Kimi K2.5 is cheaper at $0.60 input / $3.00 output per million tokens (median across 36 API providers). Qwen3.5 397B-A17B costs $0.60 input / $3.60 output per million tokens (official Alibaba API price); Grok 4.20 (Non-Reasoning) costs $1.25 input / $2.50 output per million tokens (official xAI API price). At a typical mix of three input tokens to one output token, that is $1.20 per million tokens for Kimi K2.5 versus $1.35 for Qwen3.5 397B-A17B (1.1× as much) and $1.56 for Grok 4.20 (Non-Reasoning) (1.3× as much).

Which scores higher on benchmarks?

There is no independent benchmark that covers all three models yet. Qwen3.5 397B-A17B has an ECI of 148.0, Grok 4.20 (Non-Reasoning) has not been scored yet and Kimi K2.5 has an ECI of 148.8.

Which is better for coding?

There are no published SWE-bench Verified results for Qwen3.5 397B-A17B and Grok 4.20 (Non-Reasoning) 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?

Grok 4.20 (Non-Reasoning) has the largest context window at 1,000,000 tokens, against 262,144 for Qwen3.5 397B-A17B and 262,144 for Kimi K2.5. Maximum output per response: Qwen3.5 397B-A17B up to 65,536, Grok 4.20 (Non-Reasoning) up to 30,000, Kimi K2.5 up to 262,144 tokens.

Which can read images, PDFs, audio or video?

Qwen3.5 397B-A17B accepts text, images and video; Grok 4.20 (Non-Reasoning) accepts text, images and PDFs; Kimi K2.5 accepts text, images and video. They handle the same number of input types.

Are any of these open source?

Qwen3.5 397B-A17B and Kimi K2.5 publishes its weights and can be self-hosted; Grok 4.20 (Non-Reasoning) is proprietary.

Which is newer?

Grok 4.20 (Non-Reasoning) is the newest, released Mar 9, 2026. Qwen3.5 397B-A17B came out Feb 15, 2026; Kimi K2.5 came out Jan 27, 2026. Knowledge cutoff: Kimi K2.5 Jan 2025.

How do you decide the winner?

This rating uses price (50%, blended price per million tokens (3 input : 1 output), log scale); inputs & features (30%, image, PDF, audio and video input, tool calling, structured output and reasoning); context window (20%, 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 18, 2026.