Skip to contentT/TTokens or Towers

Compare

LeaderboardEvery tracked model, ranked and priced side by sideBenchmarksWhat each test measures and who leads itPricesAPI, media, and hardware rates with a cost calculator

Plan

Stack plannerSix questions to a priced local and hosted stackShortlistOur picks for subscriptions, APIs, and hardwareEnterpriseSeat plans against self-hosting for a whole team

Build

HardwareGPUs and workstations, and which models they fitHostsWhere to rent open-weight inference, and for how muchHarnessesCoding agents and the models they drive

Lab

FindingsResults from our own evaluationsScanAudit a repo’s AI usage in the browserMethodWhere the numbers come from and how we check them
Data checked September 25, 2026
Data checked September 25, 2026
Plan a stack
T/TTokens or Towers

An independent guide to AI models, what they cost, and the hardware to run them. No account, no API key, no sponsors.

Compare

  • Leaderboard
  • Benchmarks
  • Prices

Plan

  • Stack planner
  • Shortlist
  • Enterprise

Build

  • Hardware
  • Hosts
  • Harnesses

Lab

  • Findings
  • Scan
  • Method
Data checked September 25, 2026 · Every number links to its sourceRSS · Source on GitHub
Leaderboard/Google
Google

Google · Gemma 4

DiffusionGemma 26B A4B

Open weightsPreviewApache-2.0

Experimental open-weights text-diffusion variant of Gemma 4 26B A4B that trades quality for much faster generation (1,100+ tokens/s claimed on H100).

AA Intelligence
9.5#101 of 142
Price
Self-host
Output speed
—
Context window
256K tokens

Composite indexes

  • Artificial Analysis Intelligence Index#101 of 142
    10independentv4.3.2, defaultArtificial Analysis: DiffusionGemma 26B A4B ↗

    Leader: Claude Opus 5.5 at 58

Coding

  • Terminal-Bench 2.x#89 of 103
    12.4%independentAA run (Terminal-Bench 2.1), defaultArtificial Analysis: DiffusionGemma 26B A4B ↗

    Leader: Claude Fable 5.1 at 91.4%

  • LiveCodeBench#26 of 56
    69.1%vendorLiveCodeBench v6DiffusionGemma 26B A4B model card (Hugging Face) ↗

    Leader: Fugu at 92.9%

Agents and tool use

  • GDPval-AA#87 of 103
    308independentGDPval-AA v2.1, defaultArtificial Analysis: DiffusionGemma 26B A4B ↗

    Leader: Claude Opus 5.5 at 1846

Reasoning and knowledge

  • GPQA Diamond#105 of 142
    73.2%vendorDiffusionGemma 26B A4B model card (Hugging Face) ↗
    66.9%independentAA run, defaultArtificial Analysis: DiffusionGemma 26B A4B ↗

    Leader: GPT-6 Astra at 96.1%

  • Humanity’s Last Exam#99 of 147
    10.8%independentAA run, no tools, defaultArtificial Analysis: DiffusionGemma 26B A4B ↗

    Leader: Claude Opus 5.5 at 61.4%

  • CritPt#77 of 139
    0.3%independentAA run, defaultArtificial Analysis: DiffusionGemma 26B A4B ↗

    Leader: GPT-5.6 Sol at 32.3%

  • MMLU-Pro#8 of 14
    77.6%vendorDiffusionGemma 26B A4B model card (Hugging Face) ↗

    Leader: Nemotron 3 Ultra 550B A55B at 86.8%

  • AIME 2026#11 of 13
    69.1%vendorno toolsDiffusionGemma 26B A4B model card (Hugging Face) ↗

    Leader: ERNIE 5.1 at 99.6%

Vision and long context

  • AA Long Context Reasoning#116 of 139
    19.7%independentAA-LCR v1.1, defaultArtificial Analysis: DiffusionGemma 26B A4B ↗

    Leader: Kimi K3 at 88.7%

  • MMMU-Pro#47 of 72
    66.5%independentAA run, defaultArtificial Analysis: DiffusionGemma 26B A4B ↗

    Leader: Claude Opus 5.5 at 87.7%

Other published results

ResultScoreReported bySource
Generation speed · >1,100 per user, low batch, H100 FP8 (vendor claim)1100 tokens/svendorDiffusionGemma 26B A4B model card (Hugging Face) ↗
MMMLU81.5%vendorDiffusionGemma 26B A4B model card (Hugging Face) ↗
Codeforces Elo1429 ElovendorDiffusionGemma 26B A4B model card (Hugging Face) ↗

Specs

Released
June 10, 2026
Parameters
25.2B total, 3.8B active
Size class
10 to 40B
Input
text, image, video
Output
text
Context
256K tokens
Max output
Not published

    Will it fit your budget?

    The planner prices this model against local hardware and other APIs for your workload.

    Plan a stack →

    Compare with

    • MistralMistral Large 39.3
    • AmazonAmazon Nova Premier9.2
    • CohereNorth Mini Code 1.09.9
    • IBMGranite 4.2 3B9.1