Kimi K2.5
Moonshot AI · moe · 1000B parameters · 131,072 context
Parameters
1.0T
Context Window
128K tokens
Architecture
MoE
Best GPU
B200 NVL (pair)
Cheapest API
$0.00/M
Intelligence Brief
Kimi K2.5 is a 1000B parameter Mixture-of-Experts (256 experts, 8 active) model from Moonshot AI, featuring Grouped Query Attention (GQA) with 64 layers and 6,144 hidden dimensions. With a 131,072 token context window, it supports tools, structured output, code, math, multilingual, reasoning. The most cost-effective API deployment is via fireworks at $0.00/M output tokens. For self-hosted inference, B200 NVL (pair) delivers optimal throughput at $39858/month.
Provider pricing
5 providers · canonical: moonshot| Provider | Input $/M | Output $/M ▲ | Notes |
|---|---|---|---|
| fireworks | free | free | cheapest input · cheapest output |
| featherless | free | free | cheapest input · cheapest output |
| openrouter | $0.440 | $2.00 | — |
| moonshotcanonical | $0.600 | $2.40 | — |
| novita | $0.600 | $3.00 | — |
Prices update via the nightly pricing cron + admin approvals at /admin/ingest-queue. The leaderboard's Input/Output cells show the canonical rate above; this table shows the full spread.
Recent changes
Loading…
Related models
5 suggestions- Kimi K2.7 CodeKimi · 32.9B$3.50/M out
- Kimi K3Kimi · 32B$15.00/M out
DeepSeek R1DeepSeek R1 · 37B$2.00/M out- DeepSeek V3-0324DeepSeek V3 · 37Bfree/M out
Qwen 3 235BQwen 3 · 22Bfree/M out
Picks: same family first, then same vendor within ±2× params, then top tag-overlap matches. Price shown is the cheapest Output $/M across providers — the row's page shows the canonical anchor.
Architecture Details
Memory Requirements
BF16 Weights
2000.0 GB
FP8 Weights
1000.0 GB
INT4 Weights
500.0 GB
Fits on (multi-GPU with Tensor Parallelism)
Multi-GPU configurations use Tensor Parallelism (TP) to split model layers across GPUs. Requires NVLink or NVSwitch interconnect for optimal performance.
This model requires multi-GPU deployment. Minimum: 2x B200 NVL (pair) (360GB each) with Tensor Parallelism.
GPU Compatibility Matrix
Kimi K2.5 is compatible with 0% of GPU configurations across 41 GPUs at 3 precision levels.
GPU Recommendations
FP8 · 4 GPUs · tensorrt-llm
98/100
score
Throughput
140.0 tok/s
Latency (ITL)
7.1ms
Est. TTFT
1ms
Cost/Month
$39858
Cost/M Tokens
$108.33
FP8 · 8 GPUs · tensorrt-llm
93/100
score
Throughput
140.0 tok/s
Latency (ITL)
7.1ms
Est. TTFT
1ms
Cost/Month
$34088
Cost/M Tokens
$92.65
FP8 · 8 GPUs · tensorrt-llm
93/100
score
Throughput
140.0 tok/s
Latency (ITL)
7.1ms
Est. TTFT
1ms
Cost/Month
$34164
Cost/M Tokens
$92.86
Deployment Options
API Deployment
fireworks
$0.00/M
output tokens
Single GPU
Requires multi-GPU setup (1000 GB VRAM needed)
Multi-GPU
B200 NVL (pair) x4
140.0 tok/s
TP· $39858/mo
API Pricing Comparison
| Provider | Input $/M | Output $/M | Badges |
|---|---|---|---|
| fireworks | $0.00 | $0.00 | Cheapest |
| featherless | $0.00 | $0.00 | |
| openrouter | $0.44 | $2.00 | |
| moonshot | $0.60 | $2.40 | |
| novita | $0.60 | $3.00 |
Cost Analysis
| Provider | Input $/M | Output $/M | ~Monthly Cost |
|---|---|---|---|
| fireworksBest Value | $0.00 | $0.00 | $0 |
| featherless | $0.00 | $0.00 | $0 |
| openrouter | $0.44 | $2.00 | $12 |
| moonshot | $0.60 | $2.40 | $15 |
| novita | $0.60 | $3.00 | $18 |
Cost per 1,000 Requests
Short (500 tok)
$0.00
via fireworks
Medium (2K tok)
$0.00
via fireworks
Long (8K tok)
$0.00
via fireworks
Performance Estimates
Throughput by GPU
VRAM Breakdown (B200 NVL (pair), FP8)
Precision Impact
bf16
500.0 GB
weights/GPU
fp8
250.0 GB
weights/GPU
~140.0 tok/s
int4
125.0 GB
weights/GPU
Capabilities
Features
Supported Frameworks
Supported Precisions
Where to Deploy Kimi K2.5
Similar Models
Kimi K2.7 Code
1026.9B params · moe
Quality: 50
from $3.50/M
Kimi K3
2779.9B params · moe
Quality: 50
from $15.00/M
GPT-5.5
700B params · moe
Quality: 50
from $30.00/M
DeepSeek V3-0324
685B params · moe
Quality: 81
from $0.00/M
DeepSeek R1
671B params · moe
Quality: 88
from $2.00/M
Frequently Asked Questions
How much VRAM does Kimi K2.5 need for inference?
Kimi K2.5 requires approximately 2000.0 GB of VRAM at BF16 precision, 1000.0 GB at FP8, or 500.0 GB at INT4 quantization. Additional VRAM is needed for KV-cache (262144 bytes per token) and activations (~3.00 GB).
What is the best GPU for Kimi K2.5?
The top recommended GPU for Kimi K2.5 is the B200 NVL (pair) (x4) using FP8 precision. It achieves approximately 140.0 tokens/sec at an estimated cost of $39858/month ($108.33/M tokens). Score: 98/100.
How much does Kimi K2.5 inference cost?
Kimi K2.5 API inference starts from $0.00/M input tokens and $0.00/M output tokens. Self-hosted inference costs depend on your GPU configuration — use our ROI calculator for a detailed breakdown.