MoonshotAI: Kimi K3
moonshotai/kimi-k3Kimi K3 is a 2.8T parameter open-weight multimodal reasoning model from Moonshot AI.
Modalities
In / out per 1M$0.19 / $15
Context1M
ReleasedJul 16, 2026
Providers
29 providers serve this model upstream.
| Provider | Input /M | Output /M | Cache read /M | Latency P50 | Throughput P50 | Uptime |
|---|---|---|---|---|---|---|
| $0.28 | $9.60 | $0.13 | 0.75s | 100 tps | 100% | |
| $0.19 | $15 | $0.18 | 0.81s | 92 tps | 100% | |
| $0.15 | $7.15 | $0.07 | 1.10s | 68 tps | 100% | |
| $0.15 | $22 | $0.15 | 0.81s | 94 tps | 100% | |
| $0.23 | $15 | $0.20 | 0.87s | 76 tps | 100% | |
| $0.23 | $15 | $0.20 | 0.99s | 76 tps | 100% | |
| $3.00 | $15 | $0.30 | 0.89s | 79 tps | 100% | |
| $0.50 | $22 | $0.35 | 0.78s | 96 tps | 100% | |
| $0.50 | $22 | $0.35 | 0.72s | 93 tps | 100% | |
| $3.00 | $15 | $0.30 | 0.98s | 48 tps | 100% | |
 DigitalOcean | $2.55 | $13 | $0.26 | 1.32s | 36 tps | 100% |
| $2.55 | $13 | $0.26 | 3.75s | 26 tps | 100% | |
| $3.45 | $17 | $0.34 | 2.11s | 34 tps | 100% | |
| $3.00 | $15 | $0.30 | 8.96s | 28 tps | 99.97% | |
| $0.50 | $13 | $0.20 | 1.31s | 70 tps | 99.95% | |
| $1.28 | $12 | $0.36 | 1.45s | 67 tps | 99.95% | |
 DeepInfraus · mxfp4 | $2.85 | $14 | $0.28 | 0.74s | 51 tps | 99.86% |
| $0.23 | $11 | $0.12 | 1.80s | 64 tps | 99.84% | |
| $0.18 | $9.34 | $0.18 | 1.38s | 86 tps | 99.48% | |
 Pareto Alphaus | $3.00 | $15 | $0.30 | 4.20s | 9 tps | 99.35% |
| $3.00 | $15 | $0.30 | 2.47s | 36 tps | 98.94% | |
| $1.20 | $14 | $1.20 | 0.92s | 90 tps | 98.87% | |
 Basetenfp8 | $3.00 | $15 | $0.30 | 1.02s | 78 tps | 98.62% |
 Fireworks | $3.00 | $15 | $0.30 | 1.41s | 48 tps | 98.62% |
| Fireworks | $4.50 | $23 | $0.45 | 1.73s | 78 tps | 98.11% |
 Amazon Bedrockus-east-2 | $3.00 | $15 | $0.30 | 3.07s | 33 tps | 97.61% |
| $1.85 | $9.50 | $0.18 | 3.18s | 21 tps | 95.07% | |
| Fireworks Spot | $2.10 | $11 | $0.21 | — | — | 80.07% |
| Fireworksus | $4.50 | $23 | $0.45 | 0.84s | 28 tps | 67.09% |
Market-wide measurements, refreshed every 15 minutes. They describe who serves Kimi K3 in general, not how ProjectCOZY routes your requests.
Pricing
List rates per 1M tokens — what Kimi K3 costs pay-as-you-go elsewhere, not what you pay here.
Input
$0.19
per 1M tokens
Output
$15
per 1M tokens
Cache read
$0.18
per 1M cached tokens
Included in every pass
Kimi K3 is included in every ProjectCOZY pass at no extra cost — one flat rate, no per-token metering. The rates above are there for comparison.
Performance
Time to first token and output speed across every provider, over the last 3 days.
0.0s25.0s50.0s75.0s100.0s
Oct 29, 12 AMOct 29, 8 AMOct 29, 4 PMOct 30, 12 AMOct 30, 8 AMOct 30, 4 PMOct 31, 12 AMOct 31, 8 AMOct 31, 4 PMNov 1, 12 AMNov 1, 8 AM
Together Alibaba Cloud Int. inference.net Morph (us) Sail Research Moonshot AI inference.net (us) Amazon Bedrock (us-east-2) Fireworks (us) Sail Research (us) Fireworks Spot Fireworks Wafer (us) Relace DeepInfra (us) inference.net (us) Phala Makora inference.net Chutes Wafer Wafer Modal Baseten DigitalOcean Parasail Fireworks Pareto Alpha (us) AkashML
0 tok/s1000 tok/s2000 tok/s3000 tok/s4000 tok/s
Oct 29, 12 AMOct 29, 8 AMOct 29, 4 PMOct 30, 12 AMOct 30, 8 AMOct 30, 4 PMOct 31, 12 AMOct 31, 8 AMOct 31, 4 PMNov 1, 12 AMNov 1, 8 AM
Together Alibaba Cloud Int. inference.net Morph (us) Sail Research Moonshot AI inference.net (us) Amazon Bedrock (us-east-2) Fireworks (us) Sail Research (us) Fireworks Spot Fireworks Wafer (us) Relace DeepInfra (us) inference.net (us) Phala Makora inference.net Chutes Wafer Wafer Modal Baseten Parasail Fireworks DigitalOcean Pareto Alpha (us) AkashML
Uptime
100% of requests were served successfully over the last 3 days, across all providers.
Last 3 daysAvailability 99.79%
Oct 29, 6:00 PMNow
Availability over the last 24 hoursAvailability 99.92%
100% 98% 97% 95%
Oct 31, 6:10 PMNow
Apps
Tools that most often run Kimi K3, by tokens this week.
Activity
8643.5B tokens processed market-wide in the last month.
0200.0B400.0B600.0B800.0B
Oct 2Oct 7Oct 12Oct 17Oct 22Oct 27Nov 1
Tokens
FAQ
Which model id do I use for Kimi K3?
moonshotai/kimi-k3. Send it as the model field from any OpenAI-compatible client, against https://minuykh.app/v1.
How much context does Kimi K3 hold?
1,048,576 tokens per request, of which up to 943,718 can be the response.
What can it accept and return?
Input: text, image, video. Output: text.
What does Kimi K3 cost on a pass?
Nothing beyond the pass itself. Pay-as-you-go elsewhere it lists at $0.19 per 1M input tokens and $15 per 1M output tokens.
Who built Kimi K3?
MoonshotAI, in the Kimi family.
Explore
Other models your pass can call.
