inclusionAI · Chat and reasoning · 127B parameters · 5.1B active · mit
Ling 3.0 flash
Needs 153 GB at FP8 / INT8. The smallest real configuration today is 8× Tesla V100 32GB at Vast.ai, $0.54 per hour, about $391 a month at 730 hours.
Parameters127B5.1B active per token
Context256k tokensmaximum published
Checkpoint255 GBBF16, as published
Licensemitcheck before commercial use
Released—documented date
Smallest configuration · FP8 / INT8
8× Tesla V100 32GB256 GB VRAM · 60% used
- Per hour
- $0.54
- Per month
- $391
- Cheapest at
Vast.ai
Memory · 3 precisions · 5 context lengthsHow much VRAM it needs
Weights plus runtime overhead, then the KV cache that grows with the context you serve.
| Precision | Weights | Weights + headroom | 4k context | 32k context | 128k context | 256k context |
|---|---|---|---|---|---|---|
| BF16published checkpoint | 255 GB | 306 GB | 307 GB | 308 GB | 313 GB | 319 GB |
| FP8 / INT8estimated | 127 GB | 153 GB | 153 GB | 154 GB | 159 GB | 166 GB |
| INT4 / MXFP4estimated | 76 GB | 92 GB | 92 GB | 94 GB | 98 GB | 105 GB |
| KV cacheone sequence, FP16 cache, included above | — | — | 1 GB | 2 GB | 7 GB | 13 GB |
Cross-checked on Hugging Face Oct 28, 2026: 127.5B parameters in the safetensors headers, 255.0 GB of weight files, attention layout from config.json (42 layers, latent rank 512). Method: weights as published at the native precision, verified on Hugging Face 2026-10-28; other precisions estimated at 1 (8-bit) and 0.6 (4-bit) bytes per parameter; × 1.2 runtime headroom; KV cache = 1152 bytes per token per layer × 42 full-attention layers (FP16 cache, one sequence).config.json ↗
Smallest configuration per GPU · 109 GPUs · cheapest eligible on-demand offer per GPU count, every provider on fileWhat it costs to run, GPU by GPU
The fewest GPUs of one type whose total memory fits the model at short context, priced from the cheapest eligible on-demand offer for exactly that count (45-day window, newest reading per provider), with the next providers beside it. Monthly = 730 hours.
| GPU | Smallest config | VRAM used | $ / hour | $ / month | Cheapest at |
|---|---|---|---|---|---|
| Tesla V100 32GBBest fit | 8× 32 GB256 GB total | 60% | $0.54 | $391 | Vast.aiTrust 78 · read Nov 2 · receipt · also Clore.ai $0.69, Edgevana $2.21 |
| A800 80GB PCIe | 2× 80 GB160 GB total | 95% tight | $0.96 | $702 | Vast.aiTrust 78 · read Nov 2 · receipt |
| Quadro RTX 8000 | 4× 48 GB192 GB total | 79% | $1.01 | $741 | Vast.aiTrust 78 · read Nov 2 · receipt |
| RTX 3090 | 8× 24 GB192 GB total | 79% | $1.17 | $851 |  Clore.aiTrust 49 · read Nov 2 · receipt · also Vast.ai $1.29, Lium $2.32 |
| A100 80GB | 2× 80 GB160 GB total | 95% tight | $1.20 | $878 | Vast.aiTrust 78 · read Nov 2 · receipt · also Thunder Compute $2.18, Spheron $2.38 |
| L40S | 4× 48 GB192 GB total | 79% | $1.52 | $1,110 |  LiumTrust 40 · read Nov 2 · receipt · also Vast.ai $1.87, VoltageGPU $3.04 |
| A100 40GB | 4× 40 GB160 GB total | 95% tight | $1.60 | $1,170 | Vast.aiTrust 78 · read Nov 2 · receipt · also iwinv $3.66, AtmosCompute $4.05 |
| MI300X | 1× 192 GB192 GB total | 79% | $1.65 | $1,205 |  GPUBankTrust 93 · read Oct 25 · receipt · also OpenRelay $1.99, Runpod $2.39 |
| RTX PRO 6000 | 2× 96 GB192 GB total | 79% | $1.67 | $1,218 | Vast.aiTrust 78 · read Nov 2 · receipt · also UpCloud $1.94, GPU.ai $3.38 |
| RTX A6000 | 4× 48 GB192 GB total | 79% | $1.68 | $1,226 | LiumTrust 40 · read Nov 2 · receipt · also Spheron $1.77, Vast.ai $1.82 |
| RTX PRO 5000 | 4× 48 GB192 GB total | 79% | $1.92 | $1,403 | Vast.aiTrust 78 · read Nov 2 · receipt · also Edgevana $4.60, iwinv $6.66 |
| RTX PRO 4000 | 8× 24 GB192 GB total | 79% | $1.93 | $1,406 | Vast.aiTrust 78 · read Nov 2 · receipt · also Edgevana $2.15, AtmosCompute $3.37 |
| H100 NVL | 2× 94 GB188 GB total | 81% | $2.08 | $1,521 | Clore.aiTrust 49 · read Nov 2 · receipt · also Vast.ai $5.23, AtmosCompute $8.93 |
| Gaudi 2 | 2× 96 GB192 GB total | 79% | $2.11 | $1,538 |  Cyfuture AInot scored · read Nov 2 · receipt |
| RTX 4000 Ada | 8× 20 GB160 GB total | 95% tight | $2.14 | $1,559 | Vast.aiTrust 78 · read Nov 2 · receipt |
| MI325X | 1× 256 GB256 GB total | 60% | $2.25 | $1,643 |  OpenRelaynot scored · read Nov 2 · receipt · also Cyfuture AI $3.14, DigitalOcean $3.80 |
| A40 | 4× 48 GB192 GB total | 79% | $2.30 | $1,676 | Vast.aiTrust 78 · read Nov 2 · receipt · also Denvr Cloud $2.60, Edgevana $3.01 |
| A100X | 2× 80 GB160 GB total | 95% tight | $2.56 | $1,870 | Vast.aiTrust 78 · read Nov 2 · receipt |
| RTX 6000 Ada | 4× 48 GB192 GB total | 79% | $2.60 | $1,898 |  GPU.ainot scored · read Nov 2 · receipt · also io.net $2.73, QuantaCloud $3.11 |
| L40 | 4× 48 GB192 GB total | 79% | $2.65 | $1,932 |  io.netTrust 42 · read Nov 2 · receipt · also GPU.ai $3.12, QuantaCloud $3.75 |
| RTX 4090 | 8× 24 GB192 GB total | 79% | $3.07 | $2,241 |  Edgevananot scored · read Nov 2 · receipt · also Sesterce Cloud $3.52, Vast.ai $3.63 |
| H100 | 2× 80 GB160 GB total | 95% tight | $3.42 | $2,496 |  GcoreTrust 80 · read Nov 2 · receipt · also GPU.ai $3.96, UpCloud $4.00 |
| B200 | 1× 180 GB180 GB total | 85% | $3.75 | $2,738 |  packet.ainot scored · read Nov 2 · receipt · also Beam $4.11, UpCloud $4.11 |
| RTX 5090 | 8× 32 GB256 GB total | 60% | $3.84 | $2,804 | Vast.aiTrust 78 · read Nov 2 · receipt · also Lium $4.00, Edgevana $5.28 |
| The 24 cheapest of 37 GPU types that fit. Every GPU, any precision and context, in the finder | |||||
Throughput estimates · Oct 4, 2026Tokens per second, and what a million costs
Chat and reasoning. Speed estimated from memory bandwidth and model size, never measured. The cost per 1M output tokens is an estimate at the prices read on Oct 4, 2026; it is not recomputed here.
| GPU | Cards | Precision | tok/s per user | per 1M output tokens · estimate |
|---|---|---|---|---|
| MI300X | 1 | FP8 | 127 | $0.338 |
| B200 | 1 | FP8 | 207 | $0.460 |
| H100 SXM | 2 | FP8 | 98.7 | $0.591 |
| H200 | 2 | FP8 | 104 | $0.620 |
| RTX PRO 6000 | 2 | FP8 | 64.5 | $0.732 |
| H100 PCIe | 2 | FP8 | 61.7 | $0.788 |
| L40S | 4 | FP8 | 42.8 | $0.914 |
| A100 80GB | 4 | BF16 | 60.6 | $1.07 |
cards, precision and speed estimated from memory bandwidth and model size; cost recomputed by the site from verified prices · cheapest per output at the prices of Oct 4, 2026: MI300XWhy estimates, and their limits →
Ling 3.0 flash · related