ZMIME
Comparison · 3 models · Updated Nov 22, 2026

DeepSeek V4 Flash Vision Exp vs Kolibri 1 vs Qwen3.8 Flash Next

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. DeepSeek

    DeepSeek V4 Flash Vision Exp

    Released Aug 21, 2026

    68/100
    • ECI—
    • Price$0.216 / $0.647
    • Context1M
  2. Aleph Alpha

    Kolibri 1

    Released Oct 3, 2026

    71/100
    • ECI—
    • Price$0.167 / $0.588
    • Context1.05M
  3. Alibaba (Qwen)

    Qwen3.8 Flash Next

    Released Aug 27, 2026

    65/100
    • ECI—
    • Price$0.20 / $0.50
    • Context262K
  4. DeepSeek

    DeepSeek V4 Flash Vision Exp

    Released Aug 21, 2026

    64/100
    • ECI—
    • Price$0.216 / $0.647
    • Context1M
  5. Aleph Alpha

    Kolibri 1

    Released Oct 3, 2026

    65/100
    • ECI—
    • Price$0.167 / $0.588
    • Context1.05M
  6. Alibaba (Qwen)

    Qwen3.8 Flash Next

    Released Aug 27, 2026

    43/100
    • ECI—
    • Price$0.20 / $0.50
    • Context262K
01 — Verdict

Capability evidence is incomplete

There is not enough shared capability evidence to name an overall winner. The scores (Kolibri 1 71/100, DeepSeek V4 Flash Vision Exp 68/100, Qwen3.8 Flash Next 65/100) 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 (Kolibri 1 65/100, DeepSeek V4 Flash Vision Exp 64/100, Qwen3.8 Flash Next 43/100) 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 priceKolibri 1Kolibri 1 $0.272 · Qwen3.8 Flash Next $0.275 · DeepSeek V4 Flash Vision Exp $0.323 per 1M tokens (3:1 blend)
  • Longest contextKolibri 1Kolibri 1 1,048,576 · DeepSeek V4 Flash Vision Exp 1,000,000 · Qwen3.8 Flash Next 262,144 tokens
  • Widest inputsQwen3.8 Flash NextDeepSeek V4 Flash Vision Exp: Text, Images · Kolibri 1: Text · Qwen3.8 Flash Next: Text, Images, Video
  • Self-hostingAll open weightsEvery model here can be downloaded and run on your own hardware
How the score is built
MeasureWeightDeepSeek V4 Flash Vision ExpKolibri 1Qwen3.8 Flash Next
Price71%707373
Context window29%646543
Specs & price only100%68/10071/10065/100

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

How the score is built
MeasureWeightDeepSeek V4 Flash Vision ExpKolibri 1Qwen3.8 Flash Next
Context window100%646543
Specs only100%64/10065/10043/100

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.

DeepSeek V4 Flash Vision Exp vs Kolibri 1 vs Qwen3.8 Flash Next specifications side by side
SpecificationDeepSeek V4 Flash Vision ExpDeepSeekKolibri 1Aleph AlphaQwen3.8 Flash NextAlibaba (Qwen)
Capability
Capabilities Index (ECI)Not publishedNot publishedNot published
ECI rankNot publishedNot publishedNot published
ECI confidence intervalNot publishedNot publishedNot published
ECI evaluation dateNot publishedNot publishedNot published
Documented capabilitiesExperimental multimodal DeepSeek V4 Flash model for image understanding, coding, and agentic workAleph Alpha's German and English mixture-of-experts reasoning model (78B total, 3.46B active) for chat, RAG and agentsOpen-weight experimental preview of the Qwen4 architecture: hybrid-attention MoE (125B total, 6B active) with vision encoder for coding, agent tasks, and image and video understanding
Model parametersNot publishedNot publishedNot published
Output modalitiesTextTextText
Publisher-reported benchmarks
Terminal-Bench 2.1max effort83.9scoreReported resultNot publishedNot published
NL2Repomax effort57.7scoreReported resultNot publishedNot published
CyberGymmax effort75.3scoreReported resultNot publishedNot published
DeepSWEmax effort59.3scoreReported resultNot publishedNot published
Toolathlon-Verifiedmax effort75.9scoreReported resultNot publishedNot published
DSBench-Hard · internalmax effort63.6scoreReported resultNot publishedNot published
AutomationBench · publicmax effort25.7scoreReported resultNot publishedNot published
ApexBenchPublisher-reported result36.5pass@1Reported resultNot publishedNot published
Agents' Last ExamPublisher-reported result27.3scoreReported resultNot published51.2scoreReported result
ChartographyPublisher-reported result64.3scoreReported resultNot publishedNot published
ZeroBenchPublisher-reported result35pass@5Reported resultNot publishedNot published
DeepSWE 1.1Publisher-reported resultNot publishedNot published58.7scoreReported result
SWE-Bench Pro · Qwen refined and corrected task setPublisher-reported resultNot publishedNot published62.5scoreReported result
SWE-Bench MultilingualPublisher-reported resultNot publishedNot published81scoreReported result
NL2RepoPublisher-reported resultNot publishedNot published48.1scoreReported result
CoWorkBench · internalPublisher-reported resultNot publishedNot published73.9scoreReported result
JobBenchPublisher-reported resultNot publishedNot published55.7scoreReported result
Agents' Last ExamPublisher-reported resultNot publishedNot published24.3pass@1Reported result
Toolathlon-VerifiedPublisher-reported resultNot publishedNot published73.5pass@1Reported result
IFBenchPublisher-reported resultNot publishedNot published81.3scoreReported result
GPQA DiamondPublisher-reported resultNot publishedNot published91.7scoreReported result
Humanity's Last Examwithout tools; GPT-4o judgeNot publishedNot published35.9scoreReported result
LiveCodeBench 6Publisher-reported resultNot publishedNot published91.9scoreReported result
ClawEval-MMPublisher-reported resultNot publishedNot published64.4pass@3Reported result
ClawEval-MMPublisher-reported resultNot publishedNot published60.4average scoreReported result
RecreationBench · internalPublisher-reported resultNot publishedNot published49.9scoreReported result
AndroidWorldPublisher-reported resultNot publishedNot published84.5scoreReported result
OSWorld 2.0Publisher-reported resultNot publishedNot published19.4binary completion rateReported result
OSWorld 2.0Publisher-reported resultNot publishedNot published52.3partial scoreReported result
Vision2Webgpt-5.4-2026-03-05 judgeNot publishedNot published64scoreReported result
ERQAPublisher-reported resultNot publishedNot published72.3scoreReported result
LVBenchPublisher-reported resultNot publishedNot published76.6scoreReported result
RealWorldQAPublisher-reported resultNot publishedNot published88.5scoreReported result
MathVision · corrected annotationswithout code interpreterNot publishedNot published90.6scoreReported result
MathVision · corrected annotationswith code interpreterNot publishedNot published95.7scoreReported result
CharXiv Reasoningwithout code interpreterNot publishedNot published84.6scoreReported result
CharXiv Reasoningwith code interpreterNot publishedNot published90.6scoreReported result
Price per million tokens
Input$0.216$0.167 (best)$0.20
Output$0.647$0.588$0.50 (best)
Cached inputNot publishedNot publishedNot published
Blended (3:1)$0.323$0.272 (best)$0.275
Long-context rateSame rateSame rateSame rate
Price sourceMedian of 15 providersMedian of 5 providersMedian of 5 providers
Limits
Context window1,000,000 tokens1,048,576 tokens (best)262,144 tokens
Max inputNot publishedNot publishedNot published
Max output384,000 tokens (best)Not published131,072 tokens
Inputs and features
TextYesYesYes
ImagesYesNoYes
PDFsNoNoNo
AudioNoNoNo
VideoNoNoYes
ReasoningYesYesYes
Tool callingYesYesYes
Structured outputNot publishedNot publishedNot published
File attachmentsYesNoYes
Temperature controlYesYesYes
Reasoning controlalwaysalwaysalways
Availability
WeightsOpenMITOpenApache-2.0Openqwen-community-1.0
API model IDNot publishedNot publishedNot published
API providers16 (best)56
ReleasedAug 21, 2026Oct 3, 2026Aug 27, 2026
Knowledge cutoffNot publishedJun 18, 2026Not published
Publisher documentationModel weightsModel weightsModel weights
Source metadata updatedSep 1, 2026Oct 3, 2026Aug 27, 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.

  • DeepSeek V4 Flash Vision Exp$3.45
  • Kolibri 1$2.85
  • Qwen3.8 Flash Next$3.00
04 — Questions

Which should you choose?

Which is better: DeepSeek V4 Flash Vision Exp, Kolibri 1 or Qwen3.8 Flash Next?

There is not enough shared capability evidence to name an overall winner. The scores (Kolibri 1 71/100, DeepSeek V4 Flash Vision Exp 68/100, Qwen3.8 Flash Next 65/100) 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 (Kolibri 1 65/100, DeepSeek V4 Flash Vision Exp 64/100, Qwen3.8 Flash Next 43/100) 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, DeepSeek V4 Flash Vision Exp, Kolibri 1 or Qwen3.8 Flash Next?

Kolibri 1 is cheaper at $0.167 input / $0.588 output per million tokens (median across 5 API providers). Qwen3.8 Flash Next costs $0.20 input / $0.50 output per million tokens (median across 5 API providers); DeepSeek V4 Flash Vision Exp costs $0.216 input / $0.647 output per million tokens (median across 15 API providers). At a typical mix of three input tokens to one output token, that is $0.272 per million tokens for Kolibri 1 versus $0.275 for Qwen3.8 Flash Next (1× as much) and $0.323 for DeepSeek V4 Flash Vision Exp (1.2× as much).

Which scores higher on benchmarks?

There is no independent benchmark that covers all three models yet. DeepSeek V4 Flash Vision Exp has not been scored yet, Kolibri 1 has not been scored yet and Qwen3.8 Flash Next has not been scored yet.

Which is better for coding?

There are no published SWE-bench Verified results for DeepSeek V4 Flash Vision Exp, Kolibri 1 and Qwen3.8 Flash Next 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?

Kolibri 1 has the largest context window at 1,048,576 tokens, against 1,000,000 for DeepSeek V4 Flash Vision Exp and 262,144 for Qwen3.8 Flash Next. Maximum output per response: DeepSeek V4 Flash Vision Exp up to 384,000, Qwen3.8 Flash Next up to 131,072 tokens.

Which can read images, PDFs, audio or video?

DeepSeek V4 Flash Vision Exp accepts text and images; Kolibri 1 accepts text; Qwen3.8 Flash Next accepts text, images and video. Qwen3.8 Flash Next handles the widest range of inputs.

Are any of these open source?

Yes, all three publish their weights (MIT, Apache-2.0 and qwen-community-1.0), so you can self-host them.

Which is newer?

Kolibri 1 is the newest, released Oct 3, 2026. Qwen3.8 Flash Next came out Aug 27, 2026; DeepSeek V4 Flash Vision Exp came out Aug 21, 2026. Knowledge cutoff: Kolibri 1 Jun 18, 2026.

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.