Ternary Bonsai 2 27B
- Input
- $0.075/M tok
- Output
- $0.50/M tok
- Blended
- $0.181/M tok · 3:1
- Speed
- ~84tok/sest.
- Context
- 262Ktokens
- Cost per task
- ~$0.21/taskest.
Open weights (Apache-2.0). Ternary g128 quantization of Qwen3.8-27B; PTQ1_0 GGUF is 5.95 GB (PQ2_0 7.21 GB). Scores are for the quantized weights.
Rankings
Coding
Lab-reported#14 / 137rank by quality76.5±10.4quality / 1001/7 benchmarks measuredPrice frontierWriting & chat
Lab-reported#18 / 162rank by quality76.8±9.7quality / 1001/5 benchmarks measuredPrice frontierVision
Lab-reported#26 / 98rank by quality69.5±8.7quality / 1001/6 benchmarks measuredPrice frontierHard reasoning
Lab-reported#33 / 218rank by quality73.3±9.9quality / 1001/4 benchmarks measuredPrice frontier
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
- ternary (PrismML PTQ1_0) weights
- Throughput
- ~6,364 tok/s
- many requests batched
- Price
- $0.13–1.16 /M tok
- busy → light use
- On your own machine
- RTX 5090 32GB (ternary (PrismML PTQ1_0), ~210 tok/s single-stream)
- single consumer GPU or Mac
Assumptions (5)
- TERNARY (PRISMML PTQ1_0) weights (27.36B params) + 25% KV-cache headroom ≈ 7 GB
- 1× B200 180GB at $5.98–$14.24/GPU-hour on-demand (2026-09-24)
- ~6,364 output tok/s aggregate at batch 182 (bandwidth-bound)
- Blended 3:1 input:output; prefill ~38,857 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
~#3 of 21 (estimated)~84.2%
- #1Claude Opus 596.0%
- #2Claude Sonnet 585.2%
- ~#3Ternary Bonsai 2 27BEstimated84.2%
- #3Beam 1.0 501B A23B81.6%
No published score — estimated from related benchmarks
SWE-bench Pro (Public)Estimated
~#6 of 24 (estimated)~75.2%
- #1Claude Fable 5.293.0%
- #2Claude Opus 5.589.9%
- #5Claude Opus 579.2%
- ~#6Ternary Bonsai 2 27BEstimated75.2%
- #6Intern-S2-397B68.5%
No published score — estimated from related benchmarks
DeepSWE v1.1Estimated
~#37 of 46 (estimated)~52.4%
- #1Gemini 4 Argon78.8%
- #2DeepSeek V4.1 Flash 101777.0%
- #36Nex-N2.5-Pro55.8%
- ~#37Ternary Bonsai 2 27BEstimated52.4%
- #37Qwen3.8 Max (0902)51.0%
No published score — estimated from related benchmarks
Terminal-Bench 4.0Estimated
~#16 of 102 (estimated)~38.8%
- #1Kimi K3.1 Pro67.2%
- #2Claude Sonnet 5.563.6%
- #15Qwen3.8 Max (0902)38.9%
- ~#16Ternary Bonsai 2 27BEstimated38.8%
- #16GPT-5.6 Terra35.4%
No published score — estimated from related benchmarks
LiveCodeBenchLab-reported
#4 of 2790.1%
- #1Qwen3.8-Omni-Flash92.6%
- #2Qwen3.8 Flash-Next91.9%
- #3Qwen3.8 27B90.3%
- #4Ternary Bonsai 2 27B90.1%
- #5Nemotron 3 Ultra89.0%
huggingface.co · 2026-09-29
SciCodeEstimated
~#16 of 119 (estimated)~58.1%
- #1Claude Fable 5.267.7%
- #2Claude Opus 5.566.9%
- #15Muse Spark Open58.4%
- ~#16Ternary Bonsai 2 27BEstimated58.1%
- #16GLM-5.3 NVFP458.0%
No published score — estimated from related benchmarks
AA Coding Agent IndexEstimated
~#7 of 26 (estimated)~56.4
- #1Claude Opus 5.561.7
- #2Claude Opus 559.7
- #6GPT-6 Sol56.7
- ~#7Ternary Bonsai 2 27BEstimated56.4
- #7GPT-6 Astra54.7
No published score — estimated from related benchmarks
Writing & chat
IFBenchLab-reported
#26 of 13974.0%
- #1MiniMax-M382.9%
- #2Qwen3.8 2.4T A95B82.8%
- #25LFM2.5-24B-A2B74.7%
- #26Ternary Bonsai 2 27B74.0%
- #27Command A+73.9%
huggingface.co · 2026-09-29
LMArena Text, Non-English (Elo)Estimated
~#16 of 57 (estimated)~1460 Elo
- #1Gemini 4 ArgonFew votes1512
- #2Claude Opus 5.51497
- #15MiMo-V2.6-Pro1461
- ~#16Ternary Bonsai 2 27BEstimated1460
- #16DeepSeek V4.1 Flash1459
No published score — estimated from related benchmarks
LMArena Text (Elo)Estimated
~#17 of 57 (estimated)~1473 Elo
- #1Gemini 4 Argon1525
- #2Claude Opus 5.51507
- #16DeepSeek V4.1 Flash1474
- ~#17Ternary Bonsai 2 27BEstimated1473
- #17GPT-5.6 Terra1466
No published score — estimated from related benchmarks
EQ-Bench Creative Writing v3 (Elo)Estimated
~#9 of 26 (estimated)~2005 Elo
- #1GPT-6 Astra2173
- #2Claude Fable 5.12162
- #8Grok 4.72007
- ~#9Ternary Bonsai 2 27BEstimated2005
- #9Muse Spark 1.31906
No published score — estimated from related benchmarks
Vision
MMMU-ProLab-reported
#34 of 8775.5%
- #1Claude Opus 5.587.7%
- #2GPT-6 Astra86.9%
- #32Qwen3.8 27B76.3%
- #34Ternary Bonsai 2 27B75.5%
- #35MiMo-V2.575.4%
huggingface.co · 2026-09-29
CharXiv ReasoningEstimated
~#6 of 12 (estimated)~81.6%
- #1Gemini 3.8 Flash86.2%
- #2Kimi K384.8%
- #5MiniMax-M3.1-Flash82.5%
- ~#6Ternary Bonsai 2 27BEstimated81.6%
- #6Nemotron 3.5 Super VL80.8%
No published score — estimated from related benchmarks
OmniDocBench v1.5Estimated
~#4 of 8 (estimated)~87.3%
- #1MiniMax-M391.6%
- #2Kimi K391.1%
- #2Qwen3.8 27B91.1%
- ~#4Ternary Bonsai 2 27BEstimated87.3%
- #4Gemini 3.1 Pro85.3%
No published score — estimated from related benchmarks
GDP.pdf (AA)Estimated
~#20 of 48 (estimated)~19.2%
- #1Muse Wonder Full32.6%
- #2GPT-6 Astra31.0%
- #19MiMo-V2.6-Pro19.2%
- ~#20Ternary Bonsai 2 27BEstimated19.2%
- #20Gemini 3.1 Pro17.8%
No published score — estimated from related benchmarks
ChartographyEstimated
~#9 of 25 (estimated)~31.9%
- #1Claude Fable 5.273.4%
- #2GPT-6 Astra71.0%
- #8GPT-5.6 Terra34.0%
- ~#9Ternary Bonsai 2 27BEstimated31.9%
- #9GPT-5.531.0%
No published score — estimated from related benchmarks
Hard reasoning
GPQA DiamondEstimated
~#44 of 193 (estimated)~88.7%
- #1GPT-6 Astra96.1%
- #2Gemini 3.8 Flash95.3%
- #43Nemotron 3.5 Bolt 74B-A9B88.8%
- ~#44Ternary Bonsai 2 27BEstimated88.7%
- #44DeepSeek-V4-Pro-0813-nvfp4-DSpark88.4%
No published score — estimated from related benchmarks
Humanity's Last ExamEstimated
~#41 of 199 (estimated)~37.4%
- #1Claude Fable 5.274.3%
- #2Claude Opus 5.561.4%
- #40Qwen3.8 Flash-Next38.0%
- ~#41Ternary Bonsai 2 27BEstimated37.4%
- #41Motif 337.0%
No published score — estimated from related benchmarks
AIME (latest)Lab-reported
#4 of 3395.8%
- #1GLM-5.3 NVFP499.5%
- #2Beam 1.0 501B A23B97.7%
- #3Inkling97.1%
- #4Ternary Bonsai 2 27B95.8%
- #5Ripple-65B Preview95.0%
huggingface.co · 2026-09-29
AA Intelligence IndexEstimated
~#29 of 103 (estimated)~40.8
- #1Gemini 4 Carbon59.4
- #2Claude Opus 5.557.6
- #28Gemini 3.8 Flash40.9
- ~#29Ternary Bonsai 2 27BEstimated40.8
- #29Qwen3.8 2.4T A95B39.9
No published score — estimated from related benchmarks