Nex-N2.5-Mini
- Input
- $0.025/M tok
- Output
- $0.10/M tok
- Blended
- $0.044/M tok · 3:1
- Speed
- ~110tok/sest.
- Context
- 262Ktokens
- Cost per task
- ~$0.04/taskest.
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× H200 141GB
- FP8 weights
- Throughput
- ~8,390 tok/s
- many requests batched
- Price
- $0.046–0.38 /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)
- FP8 weights (35B params, 3B active per token) + 25% KV-cache headroom ≈ 44 GB
- 1× H200 141GB at $3.59–$7.91/GPU-hour on-demand (2026-09-24)
- ~8,390 output tok/s aggregate at batch 256 (bandwidth-bound); MoE compute scales with active params
- Blended 3:1 input:output; prefill ~155,846 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
~#14 of 21 (estimated)~72.1%
- #1Claude Opus 596.0%
- #2Claude Sonnet 585.2%
- #13Claude Haiku 4.573.3%
- ~#14Nex-N2.5-MiniEstimated72.1%
- #14Nemotron 3.5 Super VL71.1%
No published score — estimated from related benchmarks
DeepSWE v1.1Lab-reported
#42 of 4536.1%
- #1Gemini 4 Argon78.8%
- #2DeepSeek V4.1 Flash 101777.0%
- #41Laguna S 2.140.4%
- #42Nex-N2.5-Mini36.1%
- #43Hy328.0%
github.com · 2026-09-08
Terminal-Bench 4.0Estimated
~#39 of 102 (estimated)~11.6%
- #1Kimi K3.1 Pro67.2%
- #2Claude Sonnet 5.563.6%
- #36Kimi K312.6%
- ~#39Nex-N2.5-MiniEstimated11.6%
- #39Qwen3.8 2.4T A95B11.1%
No published score — estimated from related benchmarks
LiveCodeBenchEstimated
~#10 of 28 (estimated)~83.8%
- #1Qwen3.8-Omni-Flash92.6%
- #2Qwen3.8 Flash-Next91.9%
- #9Nemotron 3.5 Super VL NVFP487.4%
- ~#10Nex-N2.5-MiniEstimated83.8%
- #10Ternary Bonsai 27B82.8%
No published score — estimated from related benchmarks
SciCodeEstimated
~#46 of 119 (estimated)~49.5%
- #1Claude Fable 5.267.7%
- #2Claude Opus 5.566.9%
- #45Inkling Small49.7%
- ~#46Nex-N2.5-MiniEstimated49.5%
- #46Beam 1.0 501B A23B49.2%
No published score — estimated from related benchmarks
AA Coding Agent IndexEstimated
~#22 of 26 (estimated)~38.0
- #1Claude Opus 5.561.7
- #2Claude Opus 559.7
- #21Qwen3.8 Max (0902)39.2
- ~#22Nex-N2.5-MiniEstimated38.0
- #22Claude Haiku 5.537.3
No published score — estimated from related benchmarks
Agents
τ²-bench (Telecom)Estimated
~#49 of 120 (estimated)~56.6%
- #1Step 3.7 Flash98.5%
- #2DeepSeek-V4-Pro-0813-nvfp4-DSpark98.3%
- #48Gemma 4 31B59.9%
- ~#49Nex-N2.5-MiniEstimated56.6%
- #49Claude Haiku 4.554.7%
No published score — estimated from related benchmarks
OSWorld-VerifiedLab-reported
#10 of 1471.2%
- #1Kimi K384.8%
- #2Qwen3.8 27B84.3%
- #9Gemini 3.5 Flash-Lite74.0%
- #10Nex-N2.5-Mini71.2%
- #11Muse Glimmer65.9%
github.com · 2026-09-08
OSWorld 2.0Estimated
~#4 of 5 (estimated)~48.4%
No published score — estimated from related benchmarks
BrowseCompLab-reported
#13 of 2183.4%
- #1Atria Dawn Preview92.5%
- #2GPT-6 Astra91.5%
- #12MiniMax-M383.5%
- #13Nex-N2.5-Mini83.4%
- #14Nemotron 3.5 Super VL78.0%
github.com · 2026-09-08
GDPval-AA (Elo)Estimated
~#31 of 92 (estimated)~1332 Elo
- #1Claude Opus 5.51866
- #2Muse Wonder Full1857
- #30GPT-5.51336
- ~#31Nex-N2.5-MiniEstimated1332
- #31MiniMax-M31242
No published score — estimated from related benchmarks
AutomationBench-AAEstimated
~#39 of 85 (estimated)~22.8%
- #1Claude Fable 5.276.4%
- #2Muse Wonder Full71.3%
- #38Quasar 438B23.9%
- ~#39Nex-N2.5-MiniEstimated22.8%
- #39MiniMax-M321.3%
No published score — estimated from related benchmarks
τ-Bench Banking (AA)Estimated
~#31 of 91 (estimated)~22.9%
No published score — estimated from related benchmarks
Not comparable across labs (1)
Kept for reference, not counted in rankings: these use a lab-specific task set, answer key or scoring, so scores can't be compared fairly across labs. Only published scores are shown.