K2 Horizon 375B A23B
Model ranking
Not matched to the current verified ranking snapshot.
| Provider | Price | Provider model ID | Status | Context / max output | Source |
|---|
Benchmark review · November 19, 2026. The latest verified Artificial Analysis Intelligence Index score is 18 for K2 Horizon 375B A23B, checked on November 19, 2026. Rank is among verified models in this curated catalog. Benchmark source.
K2 Horizon 375B A23B is developed by Institute of Foundation Models (IFM). This profile brings together its verified specifications, access information and independent benchmark reference so you can evaluate the model before choosing a deployment.
Architecture and context
K2 Horizon 375B A23B is an IFM text-generation model built with a sparse mixture-of-experts architecture. Its model card reports 375 billion total parameters, 23 billion active parameters and a native 524,288-token context window. The public release targets reasoning, coding and tool-using workflows.
Long-context evaluation
The large context window is relevant when a task involves multiple source files or a long collection of documents. It does not remove the need for retrieval, clear instructions or output validation. Evaluate answers against known source passages and test whether the model can distinguish quoted material from instructions.
Weights and deployment
The weights are available under Apache 2.0. The model card provides vLLM and SGLang serving recipes with K2 Horizon-specific parsing. An example deployment using eight H200 GPUs is an example configuration rather than a universal minimum. Check the exact revision: intermediate checkpoints and the final release should not be assumed to produce identical results.
Comparing K2 Horizon 375B A23B on XPU
Use the model rankings for the catalog’s current intelligence score and its checked date. Scores describe the tested configuration; a similar name or a different reasoning setting may produce different results. Compare this profile with the model rankings and review provider information before selecting an API.
Rental and token prices depend on the provider, region, routing and deployment. An empty offers table means XPU has no matching verified offer at present. It is not a zero-price quote. Check provider limits separately from the developer’s maximum context specification.
Sources and verification
K2 Horizon 375B A23B — official developer model card. Specifications reviewed on 18 October 2026. The image represents the developer’s brand, rather than a separate model-specific logo. Unknown details have been left unconfirmed instead of estimated.
Specifications
- Catalog model ID
- ifm/k2-horizon-375b-a23b
- Developer model ID / revision
- K2-Horizon-375B-A23B
- Input modalities
- textDeveloper model card ↗
- Output modalities
- textDeveloper model card ↗
- Context / input token limit
- 524,288Developer model card ↗
- Maximum output tokens
- Not verified in cited sources
- Tool calling
- SupportedDeveloper model card ↗
- Structured output
- Not verified in cited sources
- API parameters (source-specific)
- Not verified in cited sources
- Weight access
- Open weightsDeveloper model card ↗
- Weight license / access terms
- Apache 2.0Developer model card ↗
- Total parameters (billions)
- 375Developer model card ↗
- Active parameters (billions)
- 23Developer model card ↗
- Download size (GB)
- Depends on checkpoint and precision; no verified file size
- Release date
- Not verified in cited sources
- Notes
- Developer specifications may differ from hosted endpoint limits. Parameter counts follow the developer model card; weight-file metadata may report a different counting convention.
Catalog entries link to their specification and availability sources. This is a curated selection, not every model each provider offers. Best prices are the lowest collected USD tariffs, not a total-cost estimate. Token offers are selected by input price, with output from the same tariff; workload mix may change the cheapest provider. Listed variants may require batch processing, off-peak hours, or specific media settings. Price coverage varies by provider; collected tariffs include their source and observation time. Arena user-preference scores are checked automatically every six hours against the latest published Text / overall dataset. Source: Arena / LMArena; CC BY 4.0. Selection and catalog matching by XPU LIVE. They apply to the shown evaluation configuration; listed prices may use different settings. Models without verified scores are not assumed to be weaker.