Alternatives?
CheaperClaude Haiku 5.5max+20-89%$0.21Remove sampling parameters+5
OpenGLM 5.3max+9.5+23%$2.01Memory (4-bit): 228 GB → 452 GB+4
CheaperGemini 3.8 Flashhigh+10-14%*$1.24Sampling parameters are ignored+1
CheaperGPT-6 Lunamax+8.6-89%$0.068Function calling only with none · Chat Completions+3
CheaperMiMo v2.6 ProthinkingOpen+5.7-69%$0.13Memory (4-bit): 228 GB → 615 GB+4Claude Haiku 5.5max+20-89%$0.21Remove sampling parameters+5Gemini 3.8 Flashhigh+10-14%*$1.24Sampling parameters are ignored+1GPT-6 Lunamax+8.6-89%$0.068Function calling only with none · Chat Completions+3MiMo v2.6 ProthinkingOpen+5.7-69%$0.13Memory (4-bit): 228 GB → 615 GB+4GLM 5.3 FlashmaxOpen+4.8-86%$0.28Defaults to max · Self-hosted+4Claude Haiku 5.5medium+7.1-89%$0.047Remove sampling parameters+5MiMo v2.6 FlashthinkingOpen+4.3-90%$0.062Pass back reasoning_content+3GPT-6 Lunamax+8.6-89%$0.068Function calling only with none · Chat Completions+3GPT-6 Sollow+3.3+129%$0.13MiMo v2.6 ProthinkingOpen+5.7-69%$0.13Memory (4-bit): 228 GB → 615 GB+4GLM 5.3maxOpen+9.5+23%$2.01Memory (4-bit): 228 GB → 452 GB+4GLM 5.3 FlashmaxOpen+4.8-86%$0.28Defaults to max · Self-hosted+4MiMo v2.6 FlashthinkingOpen+4.3-90%$0.062Pass back reasoning_content+3
Step 5 PreviewthinkingOpen+2.4-19%$1.03Memory (4-bit): 228 GB → 363 GB+3
Qwen3.8 27BmediumOpensame-45%$1.13Context: 524K → 262K+5Run it yourself
estimateBenchmarks
in CodingWatch out 1
Training behavior may differ · Self-hosted training›
Bundled Hugging Face training may differ from native xLLM, including load-balancing loss; continue original pretraining with native xLLM checkpoints and runtime. Source ↗
FAQ
How much does K2 Horizon 375B A23B cost?›
$1 per 1M input tokens and $4 per 1M output tokens (cheapest host: Featherless).
Is K2 Horizon 375B A23B free to use?›
There is no free API; the cheapest paid price is $1 per 1M input tokens. The weights are open, so you can run it yourself.
How good is K2 Horizon 375B A23B, and what is it best at?›
#73 of 334 in Chat & Knowledge, #52 of 132 in Coding, #44 of 133 in Agents & Tool Use, #72 of 133 in Vision & Documents. Strongest at: Agents & work tasks, Reasoning, Science, Long context, Knowledge.
Is K2 Horizon 375B A23B better than Inkling Small?›
On Terminal-Bench Science 0.1, K2 Horizon 375B A23B scores 0 vs 0 — about the same.
How can I use K2 Horizon 375B A23B?›
Through the API of Featherless.
What is the context window of K2 Horizon 375B A23B?›
524K tokens.
When was K2 Horizon 375B A23B released? Is there a newer version?›
Released 2026-09-03. It is the current version in its line.
Is K2 Horizon 375B A23B open source?›
Yes, the weights are open (Apache-2.0 licence).
Can I run K2 Horizon 375B A23B locally?›
The weights are open and can be self-hosted; hardware needs are not confirmed yet.
Can K2 Horizon 375B A23B read images?›
No, it accepts text only.
What is K2 Horizon 375B A23B?›
K2 Horizon 375B A23B is an open-weights language model from Institute of Foundation Models, released Sep 2026.
What are the best alternatives to K2 Horizon 375B A23B?›
Claude Haiku 5.5 (cheaper), GLM 5.3 (open weights).