ZMIME
Comparison · 3 models · Updated Nov 22, 2026

Grok 4.20 (Non-Reasoning) vs DeepSeek V4 Pro vs MiMo-V2-Pro

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

  1. xAI

    Grok 4.20 (Non-Reasoning)

    Released Mar 9, 2026

    46/100
    • ECI—
    • Price$1.25 / $2.50
    • Context1M
  2. DeepSeek

    DeepSeek V4 Pro

    Released Apr 24, 2026

    42/100
    • ECI150.2
    • Price$1.60 / $3.33
    • Context1M
  3. Xiaomi

    MiMo-V2-Pro

    Released Mar 18, 2026

    47/100
    • ECI—
    • Price$1.00 / $3.00
    • Context1.05M
  4. xAI

    Grok 4.20 (Non-Reasoning)

    Released Mar 9, 2026

    64/100
    • ECI—
    • Price$1.25 / $2.50
    • Context1M
  5. DeepSeek

    DeepSeek V4 Pro

    Released Apr 24, 2026

    64/100
    • ECI150.2
    • Price$1.60 / $3.33
    • Context1M
  6. Xiaomi

    MiMo-V2-Pro

    Released Mar 18, 2026

    65/100
    • ECI—
    • Price$1.00 / $3.00
    • Context1.05M
01 — Verdict

Capability evidence is incomplete

There is not enough shared capability evidence to name an overall winner. The scores (MiMo-V2-Pro 47/100, Grok 4.20 (Non-Reasoning) 46/100, DeepSeek V4 Pro 42/100) compare the available specifications and price. The score weighs price 71%, context window 29%. None of the shared benchmarks cover every model here yet, so capability is left out of this verdict.There is not enough shared capability evidence to name an overall winner. The scores (MiMo-V2-Pro 65/100, Grok 4.20 (Non-Reasoning) 64/100, DeepSeek V4 Pro 64/100) compare the available specifications only. The score weighs context window 100%. 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 priceMiMo-V2-ProMiMo-V2-Pro $1.50 · Grok 4.20 (Non-Reasoning) $1.56 · DeepSeek V4 Pro $2.03 per 1M tokens (3:1 blend)
  • Longest contextMiMo-V2-ProMiMo-V2-Pro 1,048,576 · Grok 4.20 (Non-Reasoning) 1,000,000 · DeepSeek V4 Pro 1,000,000 tokens
  • Widest inputsGrok 4.20 (Non-Reasoning)Grok 4.20 (Non-Reasoning): Text, Images, PDFs · DeepSeek V4 Pro: Text · MiMo-V2-Pro: Text
  • Self-hostingDeepSeek V4 ProPublishes downloadable weights
How the score is built
MeasureWeightGrok 4.20 (Non-Reasoning)DeepSeek V4 ProMiMo-V2-Pro
Price71%393440
Context window29%646465
Specs & price only100%46/10042/10047/100

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

How the score is built
MeasureWeightGrok 4.20 (Non-Reasoning)DeepSeek V4 ProMiMo-V2-Pro
Context window100%646465
Specs only100%64/10064/10065/100

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

02 — Side by side

Every spec in one table

Highlighted cells lead their row. Unavailable specifications are marked “Not published”. Publisher results include their evaluation settings and source.

Grok 4.20 (Non-Reasoning) vs DeepSeek V4 Pro vs MiMo-V2-Pro specifications side by side
SpecificationGrok 4.20 (Non-Reasoning)xAIDeepSeek V4 ProDeepSeekMiMo-V2-ProXiaomi
Capability
Capabilities Index (ECI)Not published150.2Not published
ECI rankNot published#68 of 159Not published
ECI confidence intervalNot published148.4–152.0Not published
ECI evaluation dateNot publishedApr 24, 2026Not published
Documented capabilitiesGrok model for agentic tool use, reasoning, coding, and live assistanceOpen MoE flagship with million-token context for coding and long agent runsEarlier MiMo Pro model for multimodal agents, reasoning, and code tasks
Model parametersNot publishedNot publishedNot published
Output modalitiesTextTextText
GPQA DiamondGraduate-level science questionsNot published90.9%highNot published
FrontierMath Tiers 1–3Research-level mathematicsNot published45.3%maxNot published
OTIS Mock AIME 2024–2025Competition mathematicsNot published96.7%maxNot published
SWE-bench VerifiedFixing real GitHub issuesNot published77.6%maxNot published
SimpleQA VerifiedShort factual questionsNot published47.0%maxNot published
Publisher-reported benchmarks
SWE-Bench VerifiedPublisher-reported resultNot published80.6resolvedReported resultNot published
Artificial Analysis Coding Agent IndexhighNot published50.1average pass@1Reported resultNot published
SWE-Atlas Codebase QnAhighNot published67.8pass@1Reported resultNot published
SWE-Bench Pro · hard-aahighNot published18pass@1Reported resultNot published
Terminal-Bench 2.1highNot published64.7pass@1Reported resultNot published
MMLU-Propreview checkpoint; max effortNot published87.5EMReported resultNot published
SimpleQA-Verifiedpreview checkpoint; max effortNot published57.9pass@1Reported resultNot published
Chinese SimpleQApreview checkpoint; max effortNot published84.4pass@1Reported resultNot published
GPQA Diamondpreview checkpoint; max effortNot published90.1pass@1Reported resultNot published
Humanity's Last Exampreview checkpoint; max effort; without toolsNot published37.7pass@1Reported resultNot published
LiveCodeBenchpreview checkpoint; max effortNot published93.5pass@1Reported resultNot published
Codeforcespreview checkpoint; max effortNot published3206ratingReported resultNot published
HMMT · February 2026preview checkpoint; max effortNot published95.2pass@1Reported resultNot published
IMOAnswerBenchpreview checkpoint; max effortNot published89.8pass@1Reported resultNot published
MathArena Apexpreview checkpoint; max effortNot published38.3pass@1Reported resultNot published
MathArena Apex Shortlistpreview checkpoint; max effortNot published90.2pass@1Reported resultNot published
MRCR · 1M contextpreview checkpoint; max effortNot published83.5MMRReported resultNot published
CorpusQA · 1M contextpreview checkpoint; max effortNot published62accuracyReported resultNot published
Terminal-Bench 2.0preview checkpoint; max effortNot published67.9accuracyReported resultNot published
SWE-Bench Propreview checkpoint; max effortNot published55.4resolvedReported resultNot published
SWE-Bench Multilingualpreview checkpoint; max effortNot published76.2resolvedReported resultNot published
BrowseComppreview checkpoint; max effortNot published83.4pass@1Reported resultNot published
Humanity's Last Exampreview checkpoint; max effort; with toolsNot published48.2pass@1Reported resultNot published
MCP Atlas · publicpreview checkpoint; max effortNot published73.6pass@1Reported resultNot published
GDPval-AApreview checkpoint; max effortNot published1554EloReported resultNot published
Toolathlonpreview checkpoint; max effortNot published51.8pass@1Reported resultNot published
Price per million tokens
Input$1.25$1.60$1.00 (best)
Output$2.50 (best)$3.33$3.00
Cached input$0.20Not publishedNot published
Blended (3:1)$1.56$2.03$1.50 (best)
Long-context rateOver 200K: $2.50 / $5.00Same rateSame rate
Price sourceOfficial xAI APIMedian of 65 providersMedian of 1 providers
Limits
Context window1,000,000 tokens1,000,000 tokens1,048,576 tokens (best)
Max inputNot publishedNot publishedNot published
Max output30,000 tokens384,000 tokens (best)131,072 tokens
Inputs and features
TextYesYesYes
ImagesYesNoNo
PDFsYesNoNo
AudioNoNoNo
VideoNoNoNo
ReasoningNoYesYes
Tool callingYesYesYes
Structured outputYesNot publishedNot published
File attachmentsYesNoNo
Temperature controlYesYesYes
Reasoning controlNot publishedalwaysalways
Availability
WeightsProprietaryOpenProprietary
API model IDgrok-4.20-0309-non-reasoningNot publishedNot published
API providers1367 (best)2
ReleasedMar 9, 2026Apr 24, 2026Mar 18, 2026
Knowledge cutoffNot publishedMay 2025Dec 2024
Publisher documentationPublisher documentationModel weightsNot published
Source metadata updatedMar 9, 2026Apr 24, 2026Mar 18, 2026
Data sourcesmodels.dev · Epoch AImodels.dev · Epoch AImodels.dev · Epoch AI
Catalog checkedNov 22, 2026Nov 22, 2026Nov 22, 2026
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.

  • Grok 4.20 (Non-Reasoning)$17.50
  • DeepSeek V4 Pro$22.66
  • MiMo-V2-Pro$16.00
04 — Questions

Which should you choose?

Which is better: Grok 4.20 (Non-Reasoning), DeepSeek V4 Pro or MiMo-V2-Pro?

There is not enough shared capability evidence to name an overall winner. The scores (MiMo-V2-Pro 47/100, Grok 4.20 (Non-Reasoning) 46/100, DeepSeek V4 Pro 42/100) compare the available specifications and price. The score weighs price 71%, context window 29%. None of the shared benchmarks cover every model here yet, so capability is left out of this verdict.There is not enough shared capability evidence to name an overall winner. The scores (MiMo-V2-Pro 65/100, Grok 4.20 (Non-Reasoning) 64/100, DeepSeek V4 Pro 64/100) compare the available specifications only. The score weighs context window 100%. None of the shared benchmarks cover every model here yet, so capability is left out of this verdict.

Which is cheaper, Grok 4.20 (Non-Reasoning), DeepSeek V4 Pro or MiMo-V2-Pro?

MiMo-V2-Pro is cheaper at $1.00 input / $3.00 output per million tokens (median across 1 API provider). Grok 4.20 (Non-Reasoning) costs $1.25 input / $2.50 output per million tokens (official xAI API price); DeepSeek V4 Pro costs $1.60 input / $3.33 output per million tokens (median across 65 API providers). At a typical mix of three input tokens to one output token, that is $1.50 per million tokens for MiMo-V2-Pro versus $1.56 for Grok 4.20 (Non-Reasoning) (1× as much) and $2.03 for DeepSeek V4 Pro (1.4× as much).

Which scores higher on benchmarks?

There is no independent benchmark that covers all three models yet. Grok 4.20 (Non-Reasoning) has not been scored yet, DeepSeek V4 Pro has an ECI of 150.2 and MiMo-V2-Pro has not been scored yet.

Which is better for coding?

There are no published SWE-bench Verified results for Grok 4.20 (Non-Reasoning) and MiMo-V2-Pro 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?

MiMo-V2-Pro has the largest context window at 1,048,576 tokens, against 1,000,000 for Grok 4.20 (Non-Reasoning) and 1,000,000 for DeepSeek V4 Pro. Maximum output per response: Grok 4.20 (Non-Reasoning) up to 30,000, DeepSeek V4 Pro up to 384,000, MiMo-V2-Pro up to 131,072 tokens.

Which can read images, PDFs, audio or video?

Grok 4.20 (Non-Reasoning) accepts text, images and PDFs; DeepSeek V4 Pro accepts text; MiMo-V2-Pro accepts text. Grok 4.20 (Non-Reasoning) handles the widest range of inputs.

Are any of these open source?

DeepSeek V4 Pro publishes its weights and can be self-hosted; Grok 4.20 (Non-Reasoning) and MiMo-V2-Pro is proprietary.

Which is newer?

DeepSeek V4 Pro is the newest, released Apr 24, 2026. MiMo-V2-Pro came out Mar 18, 2026; Grok 4.20 (Non-Reasoning) came out Mar 9, 2026. Knowledge cutoff: DeepSeek V4 Pro May 2025, MiMo-V2-Pro Dec 2024.

How do you decide the winner?

This rating uses price (71%, blended price per million tokens (3 input : 1 output), log scale); context window (29%, 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 22, 2026.This rating uses context window (100%, 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 22, 2026.