ZMIME
Comparison · 3 models · Updated Nov 18, 2026

GLM-5.3 vs Grok 4.20 (Reasoning) vs Kimi K2.6

Grok 4.20 (Reasoning) comes out ahead, 77 to 74 and 73 on our weighted score, and it is the cheaper option too.

Price in rating: Excluded · Include price

  1. Z.ai (Zhipu)

    GLM-5.3

    Released Aug 14, 2026

    73/100
    • ECI157.1
    • Price$1.40 / $4.40
    • Context1M
  2. Our pick

    xAI

    Grok 4.20 (Reasoning)

    Released Mar 9, 2026

    77/100
    • ECI152.3
    • Price$1.25 / $2.50
    • Context1M
  3. Moonshot AI

    Kimi K2.6

    Released Apr 20, 2026

    74/100
    • ECI151.6
    • Price$0.95 / $4.00
    • Context262K
01 — Verdict

Grok 4.20 (Reasoning) is our pick

Grok 4.20 (Reasoning) is the better all-round choice, scoring 77/100 against Kimi K2.6 (74) and GLM-5.3 (73). GLM-5.3 wins on capability. The score weighs capability 67%, inputs & features 20%, context window 13%.

  • CapabilityGLM-5.3Capabilities Index (ECI): GLM-5.3 157.1 · Grok 4.20 (Reasoning) 152.3 · Kimi K2.6 151.6
  • Lowest priceGrok 4.20 (Reasoning)Grok 4.20 (Reasoning) $1.56 · Kimi K2.6 $1.71 · GLM-5.3 $2.15 per 1M tokens (3:1 blend)
  • Longest contextGLM-5.3 and Grok 4.20 (Reasoning)GLM-5.3 1,000,000 · Grok 4.20 (Reasoning) 1,000,000 · Kimi K2.6 262,144 tokens
  • Widest inputsGrok 4.20 (Reasoning) and Kimi K2.6GLM-5.3: Text · Grok 4.20 (Reasoning): Text, Images, PDFs · Kimi K2.6: Text, Images, Video
  • Self-hostingGLM-5.3 and Kimi K2.6Publishes downloadable weights
How the score is built
MeasureWeightGLM-5.3Grok 4.20 (Reasoning)Kimi K2.6
CapabilityCapabilities Index (ECI)67%847979
Inputs & features20%458080
Context window13%646443
Overall100%73/10077/10074/100
02 — Side by side

Every spec in one table

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

GLM-5.3 vs Grok 4.20 (Reasoning) vs Kimi K2.6 specifications side by side
SpecificationGLM-5.3Z.ai (Zhipu)Grok 4.20 (Reasoning)xAIKimi K2.6Moonshot AI
Capability
Capabilities Index (ECI)157.1 (best)152.3151.6
ECI rank#33 of 158 (best)#59 of 158#61 of 158
GPQA DiamondGraduate-level science questions90.9% (best)89.3%90.8%
FrontierMath Tiers 1–3Research-level mathematics68.8% (best)44.9%57.2%
OTIS Mock AIME 2024–2025Competition mathematics91.1%92.2%96.1% (best)
SWE-bench VerifiedFixing real GitHub issues——76.7%
SimpleQA VerifiedShort factual questions41.0% (best)30.2%34.9%
Price per million tokens
Input$1.40$1.25$0.95 (best)
Output$4.40$2.50 (best)$4.00
Cached input$0.26$0.20$0.16 (best)
Blended (3:1)$2.15$1.56 (best)$1.71
Long-context rateSame rateOver 200K: $2.50 / $5.00Same rate
Price sourceOfficial Z.AI APIOfficial xAI APIOfficial Moonshot AI API
Limits
Context window1,000,000 tokens (best)1,000,000 tokens (best)262,144 tokens
Max output131,072 tokens30,000 tokens262,144 tokens (best)
Inputs and features
TextYesYesYes
ImagesNoYesYes
PDFsNoYesNo
AudioNoNoNo
VideoNoNoYes
ReasoningYeslow · high · maxYesYes
Tool callingYesYesYes
Structured outputYesYesYes
Availability
WeightsOpenProprietaryOpen
API model IDglm-5.3grok-4.20-0309-reasoningkimi-k2.6
API providers123 (best)1464
ReleasedAug 14, 2026Mar 9, 2026Apr 20, 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.

  • GLM-5.3$22.80
  • Grok 4.20 (Reasoning)$17.50
  • Kimi K2.6$17.50
04 — Questions

Which should you choose?

Which is better: GLM-5.3, Grok 4.20 (Reasoning) or Kimi K2.6?

Grok 4.20 (Reasoning) is the better all-round choice, scoring 77/100 against Kimi K2.6 (74) and GLM-5.3 (73). GLM-5.3 wins on capability. The score weighs capability 67%, inputs & features 20%, context window 13%.

Which is cheaper, GLM-5.3, Grok 4.20 (Reasoning) or Kimi K2.6?

Grok 4.20 (Reasoning) is cheaper at $1.25 input / $2.50 output per million tokens (official xAI API price). Kimi K2.6 costs $0.95 input / $4.00 output per million tokens (official Moonshot AI API price); GLM-5.3 costs $1.40 input / $4.40 output per million tokens (official Z.AI API price). At a typical mix of three input tokens to one output token, that is $1.56 per million tokens for Grok 4.20 (Reasoning) versus $1.71 for Kimi K2.6 (1.1× as much) and $2.15 for GLM-5.3 (1.4× as much).

Which scores higher on benchmarks?

GLM-5.3 scores higher on the Capabilities Index (ECI): GLM-5.3 157.1 (#33 of 158), Grok 4.20 (Reasoning) 152.3 (#59 of 158) and Kimi K2.6 151.6 (#61 of 158). Their confidence ranges do not overlap (155.0–159.3 vs 150.6–154.1), so the gap is a real one. On individual benchmarks: GPQA Diamond — GLM-5.3 90.9%, Kimi K2.6 90.8%, Grok 4.20 (Reasoning) 89.3%; FrontierMath Tiers 1–3 — GLM-5.3 68.8%, Kimi K2.6 57.2%, Grok 4.20 (Reasoning) 44.9%; OTIS Mock AIME 2024–2025 — Kimi K2.6 96.1%, Grok 4.20 (Reasoning) 92.2%, GLM-5.3 91.1%; SimpleQA Verified — GLM-5.3 41.0%, Kimi K2.6 34.9%, Grok 4.20 (Reasoning) 30.2%.

Which is better for coding?

There are no published SWE-bench Verified results for GLM-5.3 and Grok 4.20 (Reasoning) yet, so there is no like-for-like coding score. On overall capability, GLM-5.3 leads, which tends to carry over to coding, but test on your own codebase. All three support tool calling for agent workflows.

Which has the bigger context window?

GLM-5.3 and Grok 4.20 (Reasoning) have the largest context windows (1,000,000 and 1,000,000 tokens), against 262,144 for Kimi K2.6. Maximum output per response: GLM-5.3 up to 131,072, Grok 4.20 (Reasoning) up to 30,000, Kimi K2.6 up to 262,144 tokens.

Which can read images, PDFs, audio or video?

GLM-5.3 accepts text; Grok 4.20 (Reasoning) accepts text, images and PDFs; Kimi K2.6 accepts text, images and video. Grok 4.20 (Reasoning) handles the widest range of inputs.

Are any of these open source?

GLM-5.3 and Kimi K2.6 publishes its weights and can be self-hosted; Grok 4.20 (Reasoning) is proprietary.

Which is newer?

GLM-5.3 is the newest, released Aug 14, 2026. Kimi K2.6 came out Apr 20, 2026; Grok 4.20 (Reasoning) came out Mar 9, 2026. Knowledge cutoff: Kimi K2.6 Jan 2025.

How do you decide the winner?

This rating uses capability (67%, independent benchmark results); inputs & features (20%, image, PDF, audio and video input, tool calling, structured output and reasoning); context window (13%, 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.