The LLM benchmark aggregator.

G9v3 39A5B

Other

§ 1 · Profile

What it is good at

Bars run from 50 — the average of the models each source lists — so right of the line is above par. The middle column is the gap to whoever leads that scope.

Agents50.7−17.4—/267✱0/5

Tool use51.7−22.3—/81✱0/1

Core abilities47.9−19.5—/204✱0/3

Long context42.1−22.5—/0✱✱

Reasoning48.3−18.6—/178✱0/4

Science48−15.5—/122✱0/1

Expert reasoning41.4−19.5—/0✱✱

Coding49.1−20—/164✱0/5

Agentic coding48.6−20.1—/157✱0/4

Code generation48.3−21.7—/77✱0/2

Knowledge51.2−12.3—/138✱0/2

Factuality51.8−7.2—/0✱✱

§ 2 · Head to head

What it beats, and what beats it

The same models turn up scope after scope. Counted once: where both were placed, who finished higher. Bars run right for G9v3 39A5B, left for the other.

GPT-4o Miniwon 8 · lost 0

GPT-4.1 Nanowon 8 · lost 0

Granite 4.2 30Bwon 11 · lost 1

NVIDIA Nemotron 3 Nano 4Bwon 7 · lost 1

Llama 4 Maverick Instructwon 7 · lost 1

Granite 4.2 3Bwon 7 · lost 2

Granite 4.2 8Bwon 7 · lost 2

G9v3 3Bwon 6 · lost 2

Gemini 3.5 Flash Litewon 6 · lost 2

GPT-5.4 Nanowon 6 · lost 2

Command Awon 6 · lost 2

GPT-4.1 Miniwon 6 · lost 2

DeepSeek V3won 6 · lost 2

Qwen3.5 4Bwon 5 · lost 3

Grok 4.3won 5 · lost 3

GPT-5.4 Miniwon 5 · lost 3

Claude Haiku 4.5won 5 · lost 3

Gemma 4 12Bwon 5 · lost 3

GLM-4.6won 4 · lost 4

Gemini 3.1 Prowon 4 · lost 4

MiniMax M3won 6 · lost 6

Nemotron 3 Superwon 6 · lost 6

Inklingwon 6 · lost 6

Qwen3.5 9Bwon 4 · lost 5

Qwen3.6 35B A3Bwon 5 · lost 7

Qwen3.6 Pluswon 3 · lost 5

Kimi K2.6won 3 · lost 5

Kimi K2won 3 · lost 5

GPT-4.1won 3 · lost 5

Qwen3.7 Maxwon 3 · lost 5

Qwen3 235B A22B Instructwon 3 · lost 5

GPT-5 Miniwon 3 · lost 5

K2 Horizon 3.7Bwon 3 · lost 6

K2 Horizon 7Bwon 3 · lost 6

Nemotron 3 Ultrawon 4 · lost 8

K2 Horizon 32Bwon 4 · lost 8

Gemma 4 31Bwon 4 · lost 8

GLM-5.3 Flashwon 2 · lost 6

Claude Sonnet 4.5won 2 · lost 6

Gemini 2.5 Prowon 2 · lost 6

Gemini 3 Prowon 2 · lost 6

Qwen3.8 27Bwon 3 · lost 9

DeepSeek V4 Flashwon 2 · lost 7

Claude Opus 4.5won 2 · lost 7

O4 Miniwon 2 · lost 7

DeepSeek V4.1 Flashwon 2 · lost 7

GPT-5.6 Lunawon 2 · lost 10

K2 Horizon MoVA 36B A4Bwon 2 · lost 10

Muse Glimmer-30Bwon 2 · lost 10

Claude Sonnet 4.6won 1 · lost 7

DeepSeek V3.2won 1 · lost 7

Gemini 3.8 Flashwon 1 · lost 7

Claude Sonnet 4won 1 · lost 7

Grok 4.5won 1 · lost 7

Qwen3.8 Maxwon 1 · lost 7

Gemini 2.5 Flashwon 1 · lost 8

GPT-5.2won 1 · lost 8

GLM-5.2won 1 · lost 11

Muse Spark 1.1won 0 · lost 8

Gemini 3.6 Flashwon 0 · lost 8

Gemini 3.5 Flashwon 0 · lost 8

Claude Opus 4.8won 0 · lost 8

Claude Opus 4.7won 0 · lost 8

GPT-5.4won 0 · lost 8

Grok 4won 0 · lost 8

Gemini 3.7 Flashwon 0 · lost 8

Grok 4.6won 0 · lost 8

GPT-5.5won 0 · lost 8

Claude Fable 5.1won 0 · lost 8

Claude Fable 5won 0 · lost 8

GLM-5.3won 0 · lost 9

DeepSeek V4 Prowon 0 · lost 9

O3won 0 · lost 9

Claude Opus 5won 0 · lost 9

Kimi K3won 0 · lost 9

GPT-5.6 Solwon 0 · lost 9

Claude Sonnet 5won 0 · lost 12

K2 Horizon 375B A23Bwon 0 · lost 12

GPT-5.6 Terrawon 0 · lost 12

§ 3 · Sources

Where the numbers come from

1 publication, 9 figures. Every one links to the page it was read from.

✱✱ IFM — K2 Horizon launch ↗the model’s own publisher9 measuresread 2026-09-27

tau3-Banking 22.1%Terminal-Bench 2.1 32.6%SciCode 34%Humanity's Last Exam (without tools) 17.5%GPQA Diamond 80.5%CritPt 0.3%AA-LCR 62%AA-Omniscience Accuracy 14.9%AA-Omniscience Non-Hallucination 87%

Badge

[![PublicAI Index](https://publicai.io/model-index/badge?model=g9v3-39a5b)](https://publicai.io/model-index/m/g9v3-39a5b)