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/Cohere
Cohere

Cohere · North

North Mini Code 1.0

Open weightsGenerally availableApache-2.0

Cohere's first open agentic coding model, a 30B total / 3B active MoE aimed at local and on-device coding agents.

AA Intelligence
9.9#109 of 151
Price
Self-host
Output speed
90tok/s
Context window
256K tokens

Composite indexes

  • Artificial Analysis Intelligence Index#109 of 151
    10independentIntelligence Index v4.3; reasoningArtificial Analysis: North Mini Code ↗

    Leader: Claude Opus 5.5 at 58

Coding

  • SciCode#80 of 100
    38.8%independentreasoningArtificial Analysis: North Mini Code ↗

    Leader: Claude Opus 5.5 at 66.9%

  • Terminal-Bench 4.0#66 of 96
    0.5%independentreasoningArtificial Analysis: North Mini Code ↗

    Leader: Claude Sonnet 5.5 at 70.6%

  • Terminal-Bench 2.x#75 of 105
    35.6%independentTerminal-Bench 2.1; reasoningArtificial Analysis: North Mini Code ↗

    Leader: Claude Fable 5.1 at 91.4%

Agents and tool use

  • τ²-Bench#54 of 89
    37.4%independenttelecom; reasoningArtificial Analysis: North Mini Code ↗

    Leader: GLM-5.2 at 99.1%

  • GDPval-AA#94 of 109
    299independentreasoningArtificial Analysis: North Mini Code ↗

    Leader: Muse Wonder Full at 1857

Reasoning and knowledge

  • GPQA Diamond#89 of 144
    75.7%independentreasoningArtificial Analysis: North Mini Code ↗

    Leader: GPT-6 Astra at 96.1%

  • Humanity’s Last Exam#100 of 153
    11.1%independentno tools; reasoningArtificial Analysis: North Mini Code ↗

    Leader: Claude Sonnet 5.5 at 64.5%

  • CritPt#89 of 144
    0.3%independentreasoningArtificial Analysis: North Mini Code ↗

    Leader: Muse Wonder Full at 33%

Vision and long context

  • AA Long Context Reasoning#105 of 144
    37.3%independentreasoningArtificial Analysis: North Mini Code ↗

    Leader: Kimi K3 at 88.7%

Other published results

ResultScoreReported bySource
Terminal-Bench Hard (AA) · reasoning31.1%independentArtificial Analysis: North Mini Code ↗
IFBench · reasoning57.6%independentArtificial Analysis: North Mini Code ↗
tau2-Bench Banking (AA) · reasoning6.4%independentArtificial Analysis: North Mini Code ↗
AA-Omniscience Index · reasoning-48.6 indexindependentArtificial Analysis: North Mini Code ↗

Specs

Released
June 9, 2026
Parameters
30B total, 3B active
Size class
10 to 40B
Input
text
Output
text
Context
256K tokens
Max output
64K tokens
First token
0.39s
  • Speed: Artificial Analysis leaderboard (North Mini Code, median across providers or first-party as shown) ↗

Will it fit your budget?

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

Plan a stack →

Compare with

  • MetaLlama 4 Maverick10
  • GoogleDiffusionGemma 26B A4B9.5
  • NVIDIANemotron 3 Nano Omni 30B A3B Reasoning10.3
  • MistralMistral Large 39.3