Granite 4.0 H Small
- Input
- $0.06/M tok
- Output
- $0.25/M tok
- Blended
- $0.107/M tok · 3:1
- Speed
- 14tok/s
- Context
- 128Ktokens
Long-tail import from Artificial Analysis (2026-09-24); AA variant: Granite 4.0 H Small. Price = AA median host price.
Rankings
Self-hostingEstimated
How we estimate →What it would cost to run these open weights yourself on rented GPUs, compared with the API price above.
- Hardware
- 1× B200 180GB
- FP8 weights
- Throughput
- ~9,168 tok/s
- many requests batched
- Price
- $0.074–0.66 /M tok
- busy → light use
- On your own machine
- RTX 5090 32GB (int4, ~280 tok/s single-stream)
- single consumer GPU or Mac
Assumptions (5)
- FP8 weights (32.2B params, 9B active per token) + 25% KV-cache headroom ≈ 40 GB
- 1× B200 180GB at $5.98–$14.24/GPU-hour on-demand (2026-09-24)
- ~9,168 output tok/s aggregate at batch 256 (bandwidth-bound); MoE compute scales with active params
- Blended 3:1 input:output; prefill ~118,125 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.
Writing & chat
IFBenchVerified
#111 of 13931.5%
- #1MiniMax-M382.9%
- #2Qwen3.8 2.4T A95B82.8%
- #110NVIDIA Nemotron Nano 12B v2 VL31.9%
- #111Granite 4.0 H Small31.5%
- #112Jamba 1.7 Mini31.4%
artificialanalysis.ai · 2026-09-24
MMMLUEstimated
~last of 5 (estimated)below measured range (<81.3%)
No published score — estimated from related benchmarks
LMArena Text, Non-English (Elo)Estimated
~#53 of 57 (estimated)~1293 Elo
- #1Gemini 4 ArgonFew votes1512
- #2Claude Opus 5.51497
- #52Olmo 3.1 32B Instruct1294
- ~#53Granite 4.0 H SmallEstimated1293
- #53Ring-flash-2.0Few votes1293
No published score — estimated from related benchmarks
LMArena Text (Elo)Estimated
~#53 of 57 (estimated)~1320 Elo
- #1Gemini 4 Argon1525
- #2Claude Opus 5.51507
- #52Ring-flash-2.01321
- ~#53Granite 4.0 H SmallEstimated1320
- #53gpt-oss-20b1317
No published score — estimated from related benchmarks
EQ-Bench Creative Writing v3 (Elo)Estimated
~#22 of 26 (estimated)~1427 Elo
- #1GPT-6 Astra2173
- #2Claude Fable 5.12162
- #21Gemini 3.1 Pro1491
- ~#22Granite 4.0 H SmallEstimated1427
- #22Mistral Large 31412
No published score — estimated from related benchmarks
Hard reasoning
GPQA DiamondVerified
#166 of 19241.6%
- #1GPT-6 Astra96.1%
- #2Gemini 3.8 Flash95.3%
- #165Exaone 4.0 1.2B42.4%
- #166Granite 4.0 H Small41.6%
- #167Kimi Linear 48B A3B Instruct41.2%
artificialanalysis.ai · 2026-09-24
Humanity's Last ExamVerified
#186 of 1983.8%
- #1Claude Fable 5.274.3%
- #2Claude Opus 5.561.4%
- #185Command A4.0%
- #186Granite 4.0 H Small3.8%
- #186Jamba Reasoning 3B3.8%
artificialanalysis.ai · 2026-09-24
AIME (latest)Estimated
~#29 of 34 (estimated)~25.5%
- #1GLM-5.3 NVFP499.5%
- #2Beam 1.0 501B A23B97.7%
- #28Ministral 3 14B30.0%
- ~#29Granite 4.0 H SmallEstimated25.5%
- #29Ministral 3 3B22.0%
No published score — estimated from related benchmarks
AA Intelligence IndexEstimated
~#102 of 103 (estimated)~4.0
- #1Gemini 4 Carbon59.4
- #2Claude Opus 5.557.6
- #100NVIDIA Nemotron 3 Nano 4B4.8
- ~#102Granite 4.0 H SmallEstimated4.0
- #102K2 Horizon 0.9B3.0
No published score — estimated from related benchmarks
Agents
τ²-bench (Telecom)Verified
#87 of 11917.3%
- #1Step 3.7 Flash98.5%
- #2DeepSeek-V4-Pro-0813-nvfp4-DSpark98.3%
- #86Llama 4 Maverick17.8%
- #87Granite 4.0 H Small17.3%
- #88Qwen3 Omni 30B A3B Instruct16.4%
artificialanalysis.ai · 2026-09-24
OSWorld-VerifiedEstimated
~#13 of 15 (estimated)~51.2%
- #1Kimi K384.8%
- #2Qwen3.8 27B84.3%
- #12Nemotron 3.5 Super VL NVFP452.3%
- ~#13Granite 4.0 H SmallEstimated51.2%
- #13Nemotron 3.5 Super VL51.1%
No published score — estimated from related benchmarks
OSWorld 2.0Estimated
~#3 of 5 (estimated)~53.8%
No published score — estimated from related benchmarks
BrowseCompEstimated
~#21 of 22 (estimated)~33.3%
- #1Atria Dawn Preview92.5%
- #2GPT-6 Astra91.5%
- #20Nemotron 3 Ultra44.4%
- ~#21Granite 4.0 H SmallEstimated33.3%
- #21Mistral Small 421.3%
No published score — estimated from related benchmarks
GDPval-AA (Elo)Estimated
~#58 of 92 (estimated)~721 Elo
- #1Claude Opus 5.51866
- #2Muse Wonder Full1857
- #57Ring-2.6-1T734
- ~#58Granite 4.0 H SmallEstimated721
- #58KAT Coder Pro V2720
No published score — estimated from related benchmarks
AutomationBench-AAEstimated
~#79 of 85 (estimated)~0.3%
- #1Claude Fable 5.276.4%
- #2Muse Wonder Full71.3%
- #77Granite 4.2 3B0.4%
- ~#79Granite 4.0 H SmallEstimated0.3%
- #79gpt-oss-120b0.2%
No published score — estimated from related benchmarks
τ-Bench Banking (AA)Estimated
~#91 of 91 (estimated)~3.1%
No published score — estimated from related benchmarks