K2 Horizon 0.9B
- Input
- —
- Output
- —
- Speed
- ~140tok/sest.
- Context
- 131Ktokens
Long-tail import from Artificial Analysis (2026-09-24); AA variant: K2 Horizon 0.9B. Price = AA median host price.
Rankings
Self-hostingEstimated
How we estimate →What it would cost to run these open weights yourself on rented GPUs. No API sells this model, so this is its price on the chart.
- Hardware
- 1× H200 141GB
- BF16 weights
- Throughput
- ~19,742 tok/s
- many requests batched
- Price
- $0.021–0.18 /M tok
- busy → light use
- On your own machine
- RTX 5090 32GB (int4, 300+ tok/s single-stream)
- single consumer GPU or Mac
Assumptions (5)
- BF16 weights (1.08B params) + 25% KV-cache headroom ≈ 3 GB
- 1× H200 141GB at $3.59–$7.91/GPU-hour on-demand (2026-09-24)
- ~19,742 output tok/s aggregate at batch 256 (bandwidth-bound)
- Blended 3:1 input:output; prefill ~216,344 tok/s
- Low = 75% utilization at the low GPU price; high = 20% at the high price
Benchmark scores
~ italic, dashed = no published score yet, estimated from related benchmarks.
Coding
SWE-bench VerifiedEstimated
~#20 of 21 (estimated)~57.5%
- #1Claude Opus 596.0%
- #2Claude Sonnet 585.2%
- #19NVIDIA Nemotron-Labs Dory1.0 4B59.5%
- ~#20K2 Horizon 0.9BEstimated57.5%
- #20LFM2.5-24B-A2B51.0%
No published score — estimated from related benchmarks
SWE-bench Pro (Public)Estimated
~#23 of 24 (estimated)~45.5%
- #1Claude Fable 5.293.0%
- #2Claude Opus 5.589.9%
- #22Muse Glimmer51.2%
- ~#23K2 Horizon 0.9BEstimated45.5%
- #23Nex-N2.5-Mini43.8%
No published score — estimated from related benchmarks
DeepSWE v1.1Estimated
~#43 of 46 (estimated)~32.1%
- #1Gemini 4 Argon78.8%
- #2DeepSeek V4.1 Flash 101777.0%
- #42Nex-N2.5-Mini36.1%
- ~#43K2 Horizon 0.9BEstimated32.1%
- #43Hy328.0%
No published score — estimated from related benchmarks
Terminal-Bench 4.0Verified
#63 of 1010.0%
- #1Kimi K3.1 Pro67.2%
- #2Claude Sonnet 5.563.6%
- #63Granite 4.2 8B0.0%
- #63K2 Horizon 0.9B0.0%
- #63K2 Horizon 3.7B0.0%
artificialanalysis.ai · 2026-09-24
LiveCodeBenchEstimated
~#25 of 28 (estimated)~31.5%
- #1Qwen3.8-Omni-Flash92.6%
- #2Qwen3.8 Flash-Next91.9%
- #24Devstral Small 234.8%
- ~#25K2 Horizon 0.9BEstimated31.5%
- #25Ministral 3 8B30.3%
No published score — estimated from related benchmarks
AA Coding Agent IndexEstimated
~#25 of 26 (estimated)~28.0
- #1Claude Opus 5.561.7
- #2Claude Opus 559.7
- #24GPT-6 Luna31.1
- ~#25K2 Horizon 0.9BEstimated28.0
- #25Muse Spark Open26.3
No published score — estimated from related benchmarks
Hard reasoning
GPQA DiamondVerified
#181 of 19229.3%
- #1GPT-6 Astra96.1%
- #2Gemini 3.8 Flash95.3%
- #179Tiny Aya Global30.5%
- #181K2 Horizon 0.9B29.3%
- #182LFM2.5-VL-1.6B28.9%
artificialanalysis.ai · 2026-09-24
Humanity's Last ExamVerified
#155 of 1985.4%
- #1Claude Fable 5.274.3%
- #2Claude Opus 5.561.4%
- #150NVIDIA Nemotron Nano 12B v2 VL5.5%
- #155K2 Horizon 0.9B5.4%
- #155Ministral 3 3B5.4%
artificialanalysis.ai · 2026-09-24
AIME (latest)Estimated
~#31 of 34 (estimated)~19.2%
- #1GLM-5.3 NVFP499.5%
- #2Beam 1.0 501B A23B97.7%
- #30Llama 4 Maverick19.3%
- ~#31K2 Horizon 0.9BEstimated19.2%
- #31Falcon-H1-34B-Instruct16.7%
No published score — estimated from related benchmarks
Agents
τ²-bench (Telecom)Estimated
~#61 of 120 (estimated)~28.9%
- #1Step 3.7 Flash98.5%
- #2DeepSeek-V4-Pro-0813-nvfp4-DSpark98.3%
- #60GLM-4.6V30.7%
- ~#61K2 Horizon 0.9BEstimated28.9%
- #61NVIDIA Nemotron 3 Nano 4B28.1%
No published score — estimated from related benchmarks
OSWorld-VerifiedEstimated
~#12 of 15 (estimated)~55.6%
- #1Kimi K384.8%
- #2Qwen3.8 27B84.3%
- #11Muse Glimmer65.9%
- ~#12K2 Horizon 0.9BEstimated55.6%
- #12Nemotron 3.5 Super VL NVFP452.3%
No published score — estimated from related benchmarks
OSWorld 2.0Estimated
~last of 5 (estimated)below measured range (<47.5%)
No published score — estimated from related benchmarks
BrowseCompEstimated
~#19 of 22 (estimated)~49.3%
- #1Atria Dawn Preview92.5%
- #2GPT-6 Astra91.5%
- #18Nemotron 3.5 Bolt 74B-A9B70.0%
- ~#19K2 Horizon 0.9BEstimated49.3%
- #19Mistral Medium 3.548.6%
No published score — estimated from related benchmarks
GDPval-AA (Elo)Estimated
~#79 of 92 (estimated)~363 Elo
- #1Claude Opus 5.51866
- #2Muse Wonder Full1857
- #78Mistral Large 3406
- ~#79K2 Horizon 0.9BEstimated363
- #79Nova 2.0 Lite354
No published score — estimated from related benchmarks