ERNIE 4.5 300B A47B
- Input
- $0.28/M tok
- Output
- $1.10/M tok
- Blended
- $0.485/M tok · 3:1
- Speed
- ~79tok/sest.
- Context
- 131Ktokens
Long-tail import from Artificial Analysis (2026-09-24); AA variant: ERNIE 4.5 300B A47B. 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
- 4× B200 180GB
- FP8 weights
- Throughput
- ~10,810 tok/s
- many requests batched
- Price
- $0.28–2.49 /M tok
- busy → light use
- On your own machine
- Mac Studio M5 Ultra 512GB (int4, ~36 tok/s single-stream)
- single consumer GPU or Mac
Assumptions (5)
- FP8 weights (300B params, 47B active per token) + 25% KV-cache headroom ≈ 375 GB
- 4× B200 180GB at $5.98–$14.24/GPU-hour on-demand (2026-09-24)
- ~10,810 output tok/s aggregate at batch 309 (bandwidth-bound); MoE compute scales with active params
- Blended 3:1 input:output; prefill ~90,479 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
#92 of 13939.1%
- #1MiniMax-M382.9%
- #2Qwen3.8 2.4T A95B82.8%
- #91Olmo 3.1 32B Instruct39.2%
- #92ERNIE 4.5 300B A47B39.1%
- #93Llama 3.1 Nemotron Ultra 253B v138.2%
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
~#47 of 57 (estimated)~1332 Elo
- #1Gemini 4 ArgonFew votes1512
- #2Claude Opus 5.51497
- #46gpt-oss-120b1338
- ~#47ERNIE 4.5 300B A47BEstimated1332
- #47INTELLECT-3Few votes1327
No published score — estimated from related benchmarks
LMArena Text (Elo)Estimated
~#47 of 57 (estimated)~1354 Elo
- #1Gemini 4 Argon1525
- #2Claude Opus 5.51507
- #46INTELLECT-31356
- ~#47ERNIE 4.5 300B A47BEstimated1354
- #47gpt-oss-120b1352
No published score — estimated from related benchmarks
EQ-Bench Creative Writing v3 (Elo)Estimated
~#22 of 26 (estimated)~1467 Elo
- #1GPT-6 Astra2173
- #2Claude Fable 5.12162
- #21Gemini 3.1 Pro1491
- ~#22ERNIE 4.5 300B A47BEstimated1467
- #22Mistral Large 31412
No published score — estimated from related benchmarks
Hard reasoning
GPQA DiamondVerified
#74 of 19281.1%
- #1GPT-6 Astra96.1%
- #2Gemini 3.8 Flash95.3%
- #73GPT-5.4 nano81.7%
- #74ERNIE 4.5 300B A47B81.1%
- #74Nova 2.0 Lite81.1%
artificialanalysis.ai · 2026-09-24
Humanity's Last ExamVerified
#197 of 1983.3%
- #1Claude Fable 5.274.3%
- #2Claude Opus 5.561.4%
- #196Devstral Small 23.5%
- #197ERNIE 4.5 300B A47B3.3%
- #198Kimi Linear 48B A3B Instruct2.5%
artificialanalysis.ai · 2026-09-24
AIME (latest)Estimated
~#23 of 34 (estimated)~68.9%
- #1GLM-5.3 NVFP499.5%
- #2Beam 1.0 501B A23B97.7%
- #22Phi-4-reasoning-plus78.0%
- ~#23ERNIE 4.5 300B A47BEstimated68.9%
- #23Mistral Large 338.0%
No published score — estimated from related benchmarks
AA Intelligence IndexEstimated
~#86 of 103 (estimated)~10.6
- #1Gemini 4 Carbon59.4
- #2Claude Opus 5.557.6
- #85Trinity Large Thinking10.8
- ~#86ERNIE 4.5 300B A47BEstimated10.6
- #86North Mini Code9.9
No published score — estimated from related benchmarks
Agents
τ²-bench (Telecom)Verified
#110 of 1190.0%
- #1Step 3.7 Flash98.5%
- #2DeepSeek-V4-Pro-0813-nvfp4-DSpark98.3%
- #109EXAONE 4.0 32B4.1%
- #110ERNIE 4.5 300B A47B0.0%
- #110Kimi Linear 48B A3B Instruct0.0%
artificialanalysis.ai · 2026-09-24
OSWorld-VerifiedEstimated
~#12 of 15 (estimated)~54.5%
- #1Kimi K384.8%
- #2Qwen3.8 27B84.3%
- #11Muse Glimmer65.9%
- ~#12ERNIE 4.5 300B A47BEstimated54.5%
- #12Nemotron 3.5 Super VL NVFP452.3%
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)~37.4%
- #1Atria Dawn Preview92.5%
- #2GPT-6 Astra91.5%
- #20Nemotron 3 Ultra44.4%
- ~#21ERNIE 4.5 300B A47BEstimated37.4%
- #21Mistral Small 421.3%
No published score — estimated from related benchmarks
GDPval-AA (Elo)Estimated
~#61 of 92 (estimated)~636 Elo
- #1Claude Opus 5.51866
- #2Muse Wonder Full1857
- #60MiniCPM5-2B694
- ~#61ERNIE 4.5 300B A47BEstimated636
- #61Nemotron 3.5 Lightning624
No published score — estimated from related benchmarks
AutomationBench-AAEstimated
~#54 of 85 (estimated)~4.9%
- #1Claude Fable 5.276.4%
- #2Muse Wonder Full71.3%
- #53Inkling5.0%
- ~#54ERNIE 4.5 300B A47BEstimated4.9%
- #54Gemma 4 31B4.9%
No published score — estimated from related benchmarks
τ-Bench Banking (AA)Estimated
~#68 of 91 (estimated)~8.7%
- #1Muse Spark 1.350.5%
- #2GLM-5.350.3%
- #67MiMo-V2.58.7%
- ~#68ERNIE 4.5 300B A47BEstimated8.7%
- #68Granite 4.2 8B7.6%
No published score — estimated from related benchmarks