Request a pilot
‹ PublicAI Index

The LLM benchmark aggregator.

K2 Horizon 3.7B

Institute of Foundation Models · 3.7B

Strongest in Knowledge work (#75 of 178), weakest in General intelligence (#141 of 204). Above par in 7 of 12 scopes. Among the models it meets almost everywhere, it finishes behind GPT-5.6 Terra and Claude Sonnet 5 and ahead of Gemma 4 31B and Qwen3.6 35B A3B.

§ 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#110/268
Knowledge work50.9#75/178
Tool use53.1—/81✱
Core abilities45.9#138/204
General intelligence44.1#141/204
Reasoning51.2—/178✱
Mathematics55.6—/140✱
Science47.8—/122✱
Expert reasoning52.9—/0✱
Coding49.7—/165✱
Agentic coding51.1—/157✱
Code generation43.3—/77✱

§ 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 K2 Horizon 3.7B, left for the other.

Qwen3.5 0.8Bwon 8 · lost 0
Mercury 2.5won 8 · lost 0
Llama 4 Scout Instructwon 8 · lost 0
Mistral Medium 3.5won 9 · lost 0
Mistral Large 3won 9 · lost 0
GPT-4o Miniwon 10 · lost 0
GPT-4.1 Nanowon 10 · lost 0
Granite 4.2 3Bwon 12 · lost 0
Granite 4.2 8Bwon 10 · lost 1
GPT-5.4 Nanowon 9 · lost 1
Granite 4.2 30Bwon 9 · lost 1
Qwen3.5 4Bwon 8 · lost 1
Gemini 3.5 Flash Litewon 8 · lost 2
K2 Horizon 0.9Bwon 8 · lost 2
G9v3 3Bwon 7 · lost 2
Qwen3.6 27Bwon 7 · lost 2
Kimi K2.7 Codewon 7 · lost 2
Grok Build 0.1won 7 · lost 2
Command Awon 7 · lost 2
Qwen3.5 2Bwon 6 · lost 2
Grok 4.3won 7 · lost 3
GPT-5.4 Miniwon 7 · lost 3
Nemotron 3 Superwon 7 · lost 3
GPT-4.1 Miniwon 7 · lost 3
DeepSeek V3won 7 · lost 3
G9v3 39A5Bwon 6 · lost 3
Nemotron 3 Ultrawon 7 · lost 4
Qwen3.6 Pluswon 6 · lost 4
Kimi K2.6won 6 · lost 4
Claude Haiku 4.5won 6 · lost 4
Qwen3.5 9Bwon 6 · lost 4
MiniMax M3won 7 · lost 5
Inklingwon 7 · lost 5
GPT OSS 120Bwon 5 · lost 4
Gemma 4 12Bwon 5 · lost 4
K2 Horizon 32Bwon 5 · lost 4
Gemini 3.1 Flash Litewon 5 · lost 4
Qwen3.6 35B A3Bwon 6 · lost 5
Gemma 4 31Bwon 6 · lost 5
O3 Miniwon 4 · lost 4
DeepSeek V3.2won 5 · lost 5
Gemini 2.5 Prowon 5 · lost 5
Gemini 3.1 Prowon 5 · lost 5
Mimo V2.5 Prowon 4 · lost 5
DeepSeek R1won 4 · lost 5
Claude Sonnet 4won 4 · lost 6
Claude Opus 4won 3 · lost 5
Kimi K2won 3 · lost 6
GPT-4.1won 3 · lost 6
Muse Glimmer-30Bwon 3 · lost 6
Qwen3.8 Flash Nextwon 3 · lost 6
GLM-4.6won 3 · lost 7
Qwen3.7 Maxwon 3 · lost 7
Claude Sonnet 4.5won 3 · lost 7
GPT-5 Miniwon 3 · lost 7
Claude Opus 4.6won 2 · lost 6
GPT-6 Lunawon 2 · lost 7
Grok 4.7won 2 · lost 7
Kimi K2.5won 2 · lost 7
Gemini 3 Prowon 2 · lost 7
GLM-5.3 Flashwon 2 · lost 8
Claude Sonnet 4.6won 2 · lost 8
Gemini 3.6 Flashwon 2 · lost 8
Gemini 2.5 Flashwon 2 · lost 8
Claude Opus 4.5won 2 · lost 8
Gemini 3.5 Flashwon 2 · lost 8
O4 Miniwon 2 · lost 8
Gemini 3.8 Flashwon 2 · lost 8
Grok 4.5won 2 · lost 8
Qwen3.8 Maxwon 2 · lost 8
K2 Horizon 7Bwon 2 · lost 9
DeepSeek V4 Flashwon 2 · lost 9
DeepSeek V4 Prowon 2 · lost 9
DeepSeek V4.1 Flashwon 2 · lost 9
Kimi K3won 2 · lost 9
GPT-5.6 Lunawon 2 · lost 10
GLM-5.2won 2 · lost 10
Qwen3.8 27Bwon 2 · lost 10
Mimo V2.6 Prowon 1 · lost 7
Muse Spark 1.2won 1 · lost 8
Muse Spark 1.1won 1 · lost 9
GPT-5.2won 1 · lost 9
O3won 1 · lost 9
Gemini 3.7 Flashwon 1 · lost 9
Grok 4.6won 1 · lost 9
K2 Horizon 375B A23Bwon 1 · lost 10
GLM-5.3won 1 · lost 10
GPT-5.6 Solwon 1 · lost 10
GPT-5won 0 · lost 9
Grok 4won 0 · lost 9
GPT-6 Solwon 0 · lost 9
Claude Opus 5.5won 0 · lost 9
Muse Spark 1.3won 0 · lost 9
GPT-6 Astrawon 0 · lost 9
Claude Opus 4.8won 0 · lost 10
Claude Opus 4.7won 0 · lost 10
GPT-5.4won 0 · lost 10
GPT-5.5won 0 · lost 10
Claude Fable 5.1won 0 · lost 10
Claude Fable 5won 0 · lost 10
Claude Opus 5won 0 · lost 11
Claude Sonnet 5won 0 · lost 12
GPT-5.6 Terrawon 0 · lost 12

§ 3 · Sources

Where the numbers come from

4 publications, 11 figures. Every one links to the page it was read from.

Artificial Analysis ↗1 measureread 2026-09-26
Artificial Analysis Intelligence Index 16
GDPval-AA ↗1 measureread 2026-09-26
GDPval-AA 865
AA-Briefcase ↗1 measureread 2026-09-26
AA-Briefcase 793
tau3-Banking 17.7%Terminal-Bench 2.1 25.1%SciCode 25.9%GPQA Diamond 65.4%SWE-bench Verified 68.6%HMMT Feb 2026 70.5%HLE 12.9%BFCL v4 50.9%

Badge

PublicAI Index badge for K2 Horizon 3.7B[![PublicAI Index](https://publicai.io/model-index/badge?model=k2-horizon-3-7b)](https://publicai.io/model-index/m/k2-horizon-3-7b)

✱ Placed by a one-off publication, not a board that re-ran the model. ✱✱ marks a figure the model’s own publisher printed, which counts for less again. Scores are 0–100 on the PublicAI Index scale, 50 = the average of the models each source lists. Scores belong to their publishers.

Snapshot 2026-09-26 · the full index · JSON API