MBZUAIK2 Horizon 375B A23B
K2 Horizon 375B A23B is a MBZUAI open-weights reasoning model, released 3 Sept 2026. Its best configuration ranks #210 of 785 on the BenchLeader Index at 56.3 ±5.3, in the upper half. The ± is the point: 288 other configurations score within that range, so they and this one cannot be told apart on quality alone — price and speed are what separate them. It scores highest in long context (65) and lowest in reasoning (48). Output speed of 909 tokens per second puts it in the fastest quarter, with a first answer in 4.3 s. Last measured 3 Sept 2026.
- Blended price
- –
- Output speed
- 909 tok/s
- measured by Artificial Analysis
- First answer
- 4.29 s
- Context
- 524k
- Full answer
- 4.84 s
- median, reasoning included
- Cost per run
- –
- Released
- 3 Sept 2026
- Overall index56
- Reasoning48
- Agents & tools63
- Knowledge60
- Long context65
- Composite64
How its index has moved
19 Sept 2026 to 18 Oct 2026The index is recomputed from scratch every day, so a line moves when a new benchmark result lands, when a publisher revises a score, or when the models it is normalised against change. Early movement usually means the score is still settling.
Benchmark results
See all LLM benchmarks. One column per reasoning effort. Rank is among every configuration of every model on that benchmark. Hover a score for the run it came from.
Reasoning
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| GPQA Diamond (AA)not in index | 87.3%#100 | Artificial Analysis | |
| Humanity's Last Exam (AA)not in index | 32.0%#142 | Artificial Analysis | |
| CritPt | 4.6%#161 | Artificial Analysis |
Coding
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| SciCode (AA)not in index | 42.9%#157 | Artificial Analysis | |
| Terminal-Bench 4.0 (AA)not in index | 1.5%#122 | Artificial Analysis | |
| Terminal-Bench 2.1 (AA)not in index | 71.9%#70 | Artificial Analysis | |
| FrontierCode 1.1 (AA)not in index | 32.1%#49 | Artificial Analysis |
Agents & tools
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| GDPval-AA v2.1not in index | 42.9%#57 | Artificial Analysis | |
| τ³-Banking (AA) | 34.2%#54 | Artificial Analysis | |
| AutomationBenchnot in index | 37.2%#102 | Artificial Analysis | |
| GDP.pdf (AA)not in index | 7.4%#148 | Artificial Analysis | |
| AA-Briefcase v1.1not in index | 1294#36 | Artificial Analysis | |
| Terminal-Bench-Science 0.1 (AA)not in index | 0.0%#72 | Artificial Analysis |
Knowledge
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| AA-Omniscience | -3.0#150 | Artificial Analysis | |
| AA-Omniscience: accuracynot in index | 18.2%#343 | Artificial Analysis | |
| AA-Omniscience: non-hallucinationnot in index | 74.2%#29 | Artificial Analysis |
Long context
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| AA-LCR | 80.0%#78 | Artificial Analysis | |
| MLCRnot in index | 12.2%#96 | Artificial Analysis |
Composite
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| AA Intelligence Index v4.3.2 | 30.5#118 | Artificial Analysis |
What a task costs
Estimates from list price, output speed and time to first answer for the best configuration. “With caching” assumes three-quarters of the input is served from the prompt cache. Reasoning tokens are not modelled.
| Workload | Tokens in / out | Cost | With caching | Time |
|---|---|---|---|---|
| Chat reply | 400 / 300 | – | – | 4.6 s |
| Summarize a 30-page report | 12,000 / 600 | – | – | 4.9 s |
| Code edit | 6,000 / 1,500 | – | – | 5.9 s |
| Agentic coding session | 60,000 / 4,000 | – | – | 8.7 s |
| Structured extraction | 2,000 / 200 | – | – | 4.5 s |
See also
Questions people ask
- What are K2 Horizon 375B A23B's benchmark scores?
- As of 25 Oct 2026, K2 Horizon 375B A23B scores 56.3 ±5.3 on the BenchLeader Index, strongest in long context (65) and weakest in reasoning (48). That rests on 5 independent benchmark results, and every score on this page links to its publisher.
- How does K2 Horizon 375B A23B rank?
- K2 Horizon 375B A23B is #210 of 785 ranked configurations on 25 Oct 2026. 288 other configurations score within its margin, so the exact position matters less than the band. The leader is Claude Opus 5.5 (max effort) at 72.6.
- When was K2 Horizon 375B A23B released?
- MBZUAI released K2 Horizon 375B A23B on 3 Sept 2026.
Data as of 25 Oct 2026. Compare with another model.
Cite as: BenchLeader, “K2 Horizon 375B A23B: benchmarks, pricing, speed and rank”, data as of 25 Oct 2026, https://www.benchleader.com/models/k2-horizon-375b-a23b.